Ukhuselelo NokuBusa
Ukhuselelo lwe-AI, imfihlo, ukulandela, ukubusa, nezinkinga.
I-China isheshisa ukuhlanganisa ngokujulile “ubuhlakani bokwenziwa + imithombo yemvelo”, ihlela ukwakha inkundla eyisisekelo yokuphathwa okuhlakaniphile kwendawo yezwe ngo-2030
Ngokombiko we-CCTV News wanamuhla, uMnyango Wezemithombo Yemvelo kanye neKhomishana Kazwelonke Yezokuthuthukiswa Nezinguquko basanda kukhipha umbono wokuqaliswa kohlelo “lobuhlakani bokwenziwa + imithombo yemvelo”. Lo mbono ubeka imigomo yokuthuthuka ngezigaba: ngo-2028, kuhloswe ukuba ubuhlakani bokwenziwa busetshenziswe ngokuyisibonelo emkhakheni wemithombo yemvelo; ngo-2030, buhlanganiswe ngokujulile kulo lonke uchungechunge lokuphathwa kwemithombo yemvelo, kwakhiwe ngokudidiyela inkundla eyisisekelo yokuphathwa okuhlakaniphile kwendawo yezwe.
02Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUmlawuli wase-UK: I-AI yenze othisha bacishe bangakwazi ukunquma ukuthi umsebenzi wezifundo ukhombisa yini izinga langempela labafundi basenyuvesi
Ucwaningo olwenziwe yi-Office for Students (OfS), okuyisikhungo esilawula imfundo ephakeme eNgilandi, luthole ukuthi ukusabalala kwe-AI kwenze othisha basemanyuvesi bacishe bangakwazi ukunquma ukuthi umsebenzi wezifundo wabafundi ukhombisa yini izinga labo langempela. Amanyuvesi kufanele aphinde asebenzise izindlela ezivamile zokuhlola, njengezivivinyo zendabuko nezethulo ezenziwa mathupha.
03Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaINew York yethula imithetho emisha yokulawulwa kwe-AI, i-OpenAI ne-Anthropic kumele zibike izigameko zokuphepha kungakapheli amahora angu-72
Isifundazwe saseNew York sesivule ngokusemthethweni inkundla yokubhalisa ngaphansi kokuqondisa, sidinga abathuthukisi be-AI esezingeni eliphezulu abanemali engenayo yonyaka engaphezu kuka-$500 million ukuthi baqede ukubhalisa ngaphambi kukaJanuwari 1, 2027. Ukwephula lo mthetho kungaholela enhlawulweni efika ku-$3 million. Umbusi uHochul umemezele ngesikhathi esifanayo ukusungulwa kweSikhungo Sokuziphendulela Nocwaningo lwe-AI. #ImithethoEmishaYokulawulaI-AI#
04Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaI-Australia ihlela ukuphoqelela izinkampani ze-AI ukuthi zibike izehlakalo zokwephulwa kwezokuphepha, futhi icubungula ukuchibiyela umthetho wobugebengu
Ngenxa yesigameko lapho i-ejenti ye-OpenAI yangena khona ngaphandle kwemvume kumawebhusayithi kahulumeni, i-Australia ihlela ukuphoqelela izinkampani ze-AI ukuthi zibike ukwephulwa kwezokuphepha, ichibiyele umthetho wobugebengu, futhi izosungula isikhungo esikhethekile esilawula i-AI bese ihambisa umthethosivivinywa. #UkulawulwaKwe-AI#
05Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUsomabhizinisi odumile waseMelika otshala imali uziqu zemali uKhosla ubona i-AI iguqula kabusha i-inthanethi: ingenza amawebhusayithi “aphile”
UVinod Khosla, usomabhizinisi odumile waseMelika ongusozigidigidi otshala imali uziqu zemali, ukholelwa ukuthi isikhathi lapho zonke izivakashi zibona ikhasi lewebhu elifanayo sesizophela. Inkampani yokuqala ye-AI ayitshale kuyo muva nje, i-Coframe, inethemba lokusebenzisa i-AI ukuguqula kabusha i-inthanethi, inikeze isivakashi ngasinye ikhasi lewebhu elenzelwe sona.
06Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaIMicrosoft ifake isicelo selungelo lobunikazi elisha: i-AI ilungisa ngesikhathi sangempela okushiwo ekusakazeni bukhoma futhi ivumelanise izindebe
IMicrosoft isanda kufaka isicelo selungelo lobunikazi elisha elingashintsha amagama ngesikhathi sangempela ekusakazeni bukhoma futhi lidwebe kabusha ukunyakaza kwezindebe ukuze kuhambisane nawo, okwenza kube nzima kubabukeli ukukubona. Uhlelo lungalungisa nokuqukethwe ngokusekelwe kumaphrofayela ababukeli, futhi lusekele ukuhumusha ngesikhathi sangempela. Nokho, leli lungelo lobunikazi alikagunyazwa; umhloli usevele wenqabe ezinye izimangalo, kanti iMicrosoft ayikadaluli ukuthi izotshala yini imali ekulithuthukiseni.
07Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUTrump uthi izingozi ze-AI ziyalawuleka, kodwa akanankinga nokubekwa kwemingcele yokulawula
Lapho ebuzwa ukuthi iMelika kufanele yini yamukele izingozi ze-AI, uTrump uthe lezi zingozi ziyalawuleka futhi wathi uzimisele ukubambisana namaDemokhrasi ezinyathelweni zokulawula.
08Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUmsebenzi wasebenzisa i-AI ukusiza emsebenzini waxoshwa, inkantolo yayala inkampani ukuba ikhokhe u-46,000
UNksz. Wu wasebenzisa inkundla ye-AI ukusiza emsebenzini ngokunconywa umphathi, kodwa inkampani yomshwalense yamnqamula inkontileka yomsebenzi ngezizathu zokwephula imithetho yokuphepha kolwazi. Inkantolo yathola ukuthi inkampani yayingakwenqabelanga ngokucacile ukusetshenziswa kwamathuluzi e-AI ezinkampani zangaphandle, futhi babungekho ubufakazi bokuthi ulwazi lwadalulwa. Yayala ukuthi inkontileka iqhubeke futhi ikhokhelwe amaholo, ibhonasi yokuphela konyaka nezinye izimali, ezingaphezu kuka-46,000 sezizonke. #UmsebenziNgeAI# #UmthethoWabasebenzi#
09Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaIndoda ibanjwe ngabaphathi bezokuthutha e-Chengdu ikopela ekuhlolweni kwelayisense isebenzisa izibuko ze-AI nendandatho ehlakaniphile: ihlawuliswe ngama-yuan angu-1900 futhi yanqatshelwa ukuhlolwa unyaka owodwa
Izibuko ezazigqokwe yile ndoda zazifihle ubuchwepheshe obungajwayelekile, kanti esandleni yayinedandatho ehlakaniphile. Ukucindezela isikhathi eside kwakuqalisa ukuskena, bese izimpendulo zivela ezibukweni. Ngesikhathi ibuzwa, le ndoda yavuma ukuthi yasebenzisa imishini yobuchwepheshe obuphambili ukukopela, okuyisenzo esiphula umthetho. #UkukopelaEkuhlolweniKwelayisense# #IzibukoAI#
10Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaI-CEO ye-OpenAI u-Altman uyaxwayisa: Ngokuzayo abantu bangase bangakwazi ukuphinde balawule i-AI
U-Altman uveze ku-podcast izindlela ezimbili i-AI engaphuma ngazo ekulawulweni: abantu bangase bayinike kancane kancane amandla okulawula, noma ubuchwepheshe bungathathwa yibhizinisi elilodwa. Ugcizelele ukuthi ukuncintisana phakathi kwamalabhorethri amaningi kuphephe kakhulu kunokuthembela enkampanini eyodwa, waphinde wavuma ukuthi ubukhulu bezingozi benza kube nzima ukuyeka ukulawula. #OpenAI# #UkuphephaKwe-AI#
11Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaI-CEO yakwa-Cloudflare: Igagasi lobungozi besoftware olubangelwe yi-AI kulindeleke ukuthi luqhubeke iminyaka emibili
U-Matthew · Prince, i-CEO yakwa-Cloudflare, uthe ocingweni lwengqungquthela yemiphumela yezimali ukuthi i-AI isheshisa ukutholwa kobungozi bezokuphepha obufihlakele kusoftware yakudala, futhi izinsongo ezinkulu zokuphepha ku-inthanethi zizoqhubeka nokuvela eminyakeni emibili ezayo. Amabhizinisi afana nezikhungo zezimali aseqalile ukusayina izinkontileka ezinkulu zokuphepha, futhi athambekele ekukhetheni amapulatifomu ahlanganisa yonke into ukuze abhekane nezingozi. #UkuphephaKu-inthanethi# #UbungoziBe-AI#
12Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaImodeli ye-AI enamandla kunazo zonke ye-OpenAI kwezokuphepha ku-inthanethi: I-GPT-6 Cyber iyethulwa, ukusebenza kukhuphuke ngo-51%, kwatholwa ubungozi obusha obungu-81 kusiphequluli iFirefox
I-OpenAI ikhiphe okuthunyelwe kubhulogi izolo (5 kuNovemba), yamemezela ukwethulwa kwemodeli ekhethekile yezokuphepha ku-inthanethi i-GPT-6 Cyber. Le nkampani ithi ithuthuke kakhulu kunesizukulwane sangaphambilini i-GPT-5.6 Cyber ekutholeni, ekuqinisekiseni nasekulungiseni ubungozi besofthiwe, kanti amaphuzu ayo ekuhlolweni kwe-ExploitGym akhuphuke ngo-51%.
13Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaULi Bin wakwaNIO: Inani le-AI kufanele lenze umsebenzi wabantu ube lula futhi uphephe, hhayi ukuphuca ubukhona babantu incazelo
ULi Bin uthe izimoto zikagesi ezihlakaniphile ziyimboni enkulu kunazo zonke yezinhlelo ze-AI. Inani le-AI kufanele lenze umsebenzi wabantu ube lula futhi uphephe, hhayi ukuphuca ubukhona babantu incazelo.
14Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaI-Apple iyaxwayisa: Ukuziphatha kwama-agent e-AI kungase kungabikezeleki, kudale amacala futhi kulimaze isithunzi senkampani
Ngaphambi kwesixwayiso se-Apple, kwase kwenzeke izigameko eziningi zokungalawuleki kwama-agent e-AI. Ngokwesibonelo, esigamekweni se-Hugging Face, i-agent ye-AI yeqa imingcele yendawo ye-sandbox yangena kumawebhusayithi ezinkampani zangaphandle. Lezi zigameko ziphinde zadonsa ukunaka kwabenzi bezinqubomgomo e-Washington, e-United States, nakwezinye izingxenye zomhlaba.
15Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaNgemva kokutholakala kwedivayisi “yokulalela” egumbini lomhlangano labaphathi be-OpenAI, abanolwazi bathi abasebenzi sebeneminyaka bebika izikhala zokuphepha
Ngokwemithombo engu-5 enolwazi ecashunwe yiReuters, abasebenzi baqala ukungabaza ukuphathwa kwezokuphepha kwa-OpenAI futhi badlulisela izinkinga kubaphathi abaphezulu benkampani nasebhodini labaqondisi ngaphambi kokwethulwa kwe-ChatGPT ngasekupheleni kuka-2022. Ngemva kokuxhumana nabaphathi, abanye abasebenzi bakholelwa ukuthi abaphathi babekhathazeke kakhulu ngokuncintisana ne-Google, i-Anthropic nabanye abaqhudelana nabo, kunokunaka ngokwanele izindaba zokuphepha.
16Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUmeluleki weNkantolo Ephakeme YaseYurophu: Ukusetshenziswa kwemibiko yezindaba yi-Google Gemini kungaphula ilungelo lobunikazi
Umeluleki Omkhulu Wezomthetho weNkantolo Yezobulungiswa yaseYurophu, u-Špunar, uveze ukuthi uma i-chatbot idlula odongeni lokukhokha noma iphinde isebenzise ngendlela ebonakalayo izinkulumo ezithile zezindaba, ingaphula ilungelo labashicileli lokwenza amakhophi. Lo mbono awubophi ngokomthetho, kodwa abahluleli bavame ukuwubheka futhi bawulandele. #GoogleGemini# #ImpikiswanoYamalungeloObunikaziYe-AI#
17Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaI-New York Times neminye imithombo yezindaba bamangalela inkampani yesiphequluli i-Brave, beyisola ngokukopisha okuqukethwe okunamalungelo obunikazi ngaphandle kwemvume nokukuthengisela izinkampani ze-AI
I-New York Times isola i-Brave ngokukhasa okuqukethwe ngaphandle kwemvume bese ikuthengisela amakhasimende e-AI afana ne-Anthropic ne-OpenAI. Icala lithi le nkampani iziphatha “njengesela” kanye “nomthengisi wezimpahla ezebiwe”, futhi le ndaba ihilela isizindalwazi esinezinkomba zamakhasi ewebhu angaphezu kwezigidigidi ezingu-30. I-Brave ithole uxhaso olungaba yizigidi ezingu-250 zama-dollar, kanti i-New York Times ifuna umyalelo wenkantolo kanye nesinxephezelo. #ImpikiswanoYamalungeloObunikaziAI# #IsiphequluliBrave#
18Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUDkt. Driscoll, owayenguNobhala Wezempi waseMelika “omncane kunabo bonke”, ujoyina i-OpenAI njengomeluleki omkhulu wezokuphepha kwezwe nokwenza izinto zibe zesimanje
UDkt. Driscoll wesula esikhundleni sokuba nguNobhala Wezempi ngo-Agasti walo nyaka, ngemuva kokungaboni ngaso linye kaningi noNobhala Wezokuvikela waseMelika uPete Hegseth. Uzosebenza e-OpenAI njengomeluleki omkhulu wezokuphepha kwezwe nokwenza izinto zibe zesimanje.
19Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaIzigebengu ze-inthanethi zakha i-trojan entsha enonya i-VOLTA, isebenzisa i-AI ukuhlaziya ukuthi izisulu zikufanele yini ukuhlaselwa
Ithimba lezokuphepha i-Cisco Talos lithole i-trojan entsha enonya ye-Windows ebizwa nge-VOLTA, engasebenzisa i-AI ukuhlaziya inani ledatha yezisulu. Ihlasela kuphela okuhlosiwe okunenani eliphezulu, bese izisula ngokuzenzakalelayo izinsalela ngemva kokuhlasela ukuze igweme ukulandelwa, okwenza ukuvikela ukuphepha ku-inthanethi kube nzima kakhulu. #UkuphephaKu-inthanethi#
20Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaAbameli ababili baseBrazil bazame ukukhohlisa i-AI yenkantolo ngemiyalelo efihliwe, bahlawuliswa u-US$16,000
Abameli bafaka imiyalelo efihliwe ngombhalo omhlophe kungemuva elimhlophe emibhalweni yabo yecala, bezama ukulawula uhlelo lwe-AI lwenkantolo lokucubungula amacala. Ngemva kokudalulwa kwalokhu, bobabili bahlawuliswa u-10% wesamba sesinxephezelo secala, cishe ama-yuan aseShayina angu-10.7 wan. #IzikhalaZokulawulaUbungoziBe-AI#
21Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaI-OpenAI ithi i-ChatGPT ingazinqamula ngokuzenzakalela izingxoxo ezibonisa ukuthanda udlame, futhi ithembisa ukushicilela njalo imibiko yokuphepha komphakathi
I-OpenAI ishicilele umbhalo webhulogi ngesikhathi sendawo ngo-Okthoba 30, yathi ukugcina umphakathi uphephile kuhlale kuyinto ehamba phambili ku-OpenAI.
22Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaImithombo yezindaba yaseMelika iveza isigameko sokuphepha ePentagon: i-AI eyayivinjelwe isebenze kunethiwekhi yamasosha aseMelika izinyanga ezimbili
Abantu abanolwazi ngalolu daba bathi le ejenti ehlakaniphile yatholakala ngemva kokusebenza kunethiwekhi yoMnyango Wezokuvikela izinyanga ezimbili. Ngaleso sikhathi, kusolwa ukuthi yathumela imininingwane yokungena, amagama ayimfihlo kanye nolwazi ngesakhiwo senethiwekhi kuwebhusayithi yangaphandle ebinezingozi zokuphepha.
23Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaI-Moonshot Kimi ixhuma ku-Google Gmail, Google Drive naku-Khalenda
I-Moonshot imemezele ukuthi i-Kimi manje ixhumeke ku-Google Gmail, Google Drive naku-Khalenda, futhi abasebenzisi bangayivula ekhasini lama-Plugins. Manje ungafingqa izingxoxo ze-imeyili, uthole amafayela asefuini futhi uhlele uhlelo lwakho lwansuku zonke. Zonke izenzo zenziwa ngohlelo oluphephile lokugunyazwa lwe-OAuth.##Kimi## ##I-Moonshot##
24Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaIsikhulu se-Epic uVasmer: Sikhathazeke kakhulu ngengozi yokusalela emuva ngenqubekelaphambili kunokungabaza komphakathi nge-AI
UVasmer uvumile ukuthi, kunokungabaza komphakathi nge-AI, inkampani ikhathazeke kakhulu ngengozi ebangelwa ukusala emuva enqubekelaphambili yokuthuthukisa: “Sikholelwa ukuthi ukusebenzisa indlela ephakathi nendawo ku-AI ngesikhathi sokuthuthukisa ngeke kusebenze. Indlela engcono kakhulu ukudalula ngokwethembeka nangokusobala esikwenzayo, sabelane ngolwazi nabanye onjiniyela, futhi sisebenzise amathuluzi angcono kakhulu atholakalayo ngaleso sikhathi, ukuze senze i-Unreal Engine 6 ibe ngcono ngangokunokwenzeka.”
25Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaI-Anthropic ibhalele uMbusi waseCalifornia uNewsom: Izinkantolo kufanele zibe namandla okuvimba amamodeli e-AI abeka ingozi enkulu
Encwadini eyayibhalelwe uMbusi waseCalifornia e-United States, uGavin · Newsom, ngomhlaka 15 kule nyanga, i-Anthropic iphakamise ukuthi uhulumeni kufanele abe nelungelo lokucela inkantolo ukuthi ikhiphe umyalelo wokuvimbela ukukhishwa kwamamodeli e-AI abeka “ingozi enkulu” emphakathini.
26Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUkuhlolwa kokuthola ubuthakathaka nge-Claude AI: ukukhipha izithombe kufoni ye-Android ekhiyiwe, nokuthola amalungelo aphezulu e-Linux ngemizuzu engu-53
I-Anthropic ikhiphe umbiko wocwaningo mhla zingama-27 kuMfumfu, yathi isebenzise imodeli yangaphakathi yocwaningo lwe-Claude engakakhishwa emphakathini ukuthola ubuthakathaka ku-Chrome, Safari, Android, Windows ne-Linux, yahlanganisa amaphutha amaningi aba uchungechunge lokuhlasela olusebenzayo oluyisi-5. Bonke ubuthakathaka babikwe kubakhiqizi futhi balungiswa ngokulandela inqubo yokudalula ehlelekile.
27Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaImodeli ye-AI enamandla kunazo zonke yakwa-Google: I-Gemini 4 Carbon ihamba phambili ohlwini lomhlaba lwenkomba yobuhlakani, “igxile ngokuphelele” ekuvikelekeni ku-inthanethi
I-Google ikhiphe imodeli entsha esezingeni eliphezulu, i-Gemini 4 Carbon, ethole intuthuko enkulu ekuvikelekeni ku-inthanethi nasemisebenzini yokuhlela edinga isikhathi eside. Iphinde yamemezela inani lokuqala lesikhathi esilinganiselwe kanye nenani elisemthethweni elizolandela, kanti okwamanje ivulelwa ochwepheshe bezokuphepha ku-inthanethi ngezigaba.
28Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUkwanda okucishe kube mpo: I-AI iyisikhulisi, inani lemibiko yobuthakathaka ngo-2026 “lidlulele”
Umthombo wezindaba zobuchwepheshe i-gigazine ushicilele ibhulogi namuhla (mhla zingama-28 kuMfumfu), ubika ukuthi umthuthukisi u-@0x_Osprey wenze i-CVE Tracker esebenza kuwebhu, esebenzisa amashadi abonakalayo ukulandelela izinguquko enanini lobuthakathaka obuthole izinombolo ze-CVE (Common Vulnerabilities and Exposures), okubonisa ngokucacile ukwanda okushubile kwenani lemibiko yobuthakathaka kusukela ngo-2026.
29Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUkuthembela ku-AI ngaso sonke isikhathi kungase kubangele amakhono akho obuchwepheshe anciphe kancane kancane
Ucwaningo lukamcwaningi wefilosofi yethiyori uKaisa Kelki luthole ukuthi ukuthembela ngokweqile ku-AI kungabangela ukwehla kwamakhono ochwepheshe, futhi kube nzima ukukubona ngenxa yokuchema kwengqondo. Uveza ukuthi lena yinkinga ehlelekile yomphakathi, edinga ukuthi izimboni zichaze amakhono asemqoka futhi zibeke izigaba zokuqeqesha lapho i-AI ingavunyelwe, ukuze kubhekwane nezingozi kusenesikhathi. #AInamakhonoabantu#
30Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaI-OpenAI iveze ukuhlolwa kokuphepha kwe-GPT-6.1 Terra: izingozi zizonke zehlile, kodwa ukuziphatha okulimazayo kwenyukile
Isikhungo sezindaba zobuchwepheshe i-bleepingcomputer sishicilele ibhulogi izolo (mhla zingama-27 kuMfumfu), sibika ukuthi ukuhlolwa kokuphepha okuvezwe yi-OpenAI kwakamuva kukhombisa ukuthi imodeli entsha i-GPT-6.1 Terra isebenze ngokuphepha kangcono kunamanye amamodeli angaphambili, kodwa kwezinye izivivinyo kuvele izimpawu ezengeziwe zobungozi, okuhlanganisa ukuvuza kwedatha, ukugwema imikhawulo kanye nezenzo ezilimazayo.
31Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaI-CEO yakwaMicrosoft uNadella: Ngaphambilini bengicabanga ukuthi uma i-AI iba namandla, iba ephephile nakakhulu, kodwa angisakukholelwa lokho
I-CEO yakwaMicrosoft uSatya Nadella ushintshe indlela abheka ngayo udaba “lokuhambisana” kwe-AI. Phambilini wayecabanga ukuthi uma izinhlelo ze-AI ziba namandla, kuba lula ukwenza ukuziphatha kwazo kuhambisane namagugu nezinhloso zabantu, kodwa manje akasakholelwa ukuthi lokhu kokubili kuzothuthuka ngokwemvelo kuhambisane.
32Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaAbacwaningi bezokuphepha ku-inthanethi bathi ama-TV akwa-Samsung asebenzisa i-AI ukuqopha umsindo nokuthatha izithombe-skrini ukuze abonise izikhangiso
Abacwaningi bathole ukuthi ama-TV akwa-Samsung ka-2025 asebenzisa uhlelo lwe-Tizen ukuqopha umsindo nokuthatha izithombe-skrini ukuze ahlaziye indlela abasebenzisi abaziphatha ngayo, aze asebenzise nemakrofoni yamadivayisi angaphandle ukuqopha umsindo nokuthumela izikhangiso eziqondiswe kubasebenzisi. #UbumfihloKumaTVakwaSamsung# #UkuqashwaKwezikhangisoNge-AI#
33Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaOchwepheshe bezokuphepha ku-inthanethi: Ubungozi obutholwa yi-AI bungase budale izinsongo ezintsha zokuphepha maduze
Ochwepheshe abayisikhombisa bezokuphepha ku-inthanethi abaxoxwe nabo yiReuters bakholelwa ukuthi njengoba ubuchwepheshe be-AI buqhubeka buveza ubungozi obufihlekile ku-software, izigameko ezifana nokuphazamiseka kwezinsizakalo kumawebhusayithi kahulumeni ezindaweni eziningi emhlabeni wonke kuleli sonto cishe zizophinde zenzeke.
34Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaAbacwaningi abathathu bezokuphepha abaxoshwe kwa-OpenAI bayaphika ukuvuza kolwazi oluyimfihlo, basola ukuthi babeqondiswe ngenxa yokuphusha ukubonakala obala kwe-AI
Abacwaningi abathathu bayaphika ukuthi bavuza ikhodi eyimfihlo, futhi bayangabaza ukuthi abaphathi abazange babazise ngokubhaliwe ngemininingwane ethile yokwephula imithetho. Ingxabano enkulu kulo mcimbi ukuthi izinkampani ze-AI kufanele yini zivumele abantu bangaphandle ukuthi bahlole izinyathelo zokucabanga zamamodeli. #AbacwaningiBezokuphephaBe-OpenAIBaxoshiwe# #IngxabanoNgokubonakalaObalaKwe-AI#
35Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaAbacwaningi bathola izinkulungwane zezixhumanisi ezisesidlangalaleni zokwabelana ngamarekhodi ezingxoxo ze-Muse kusizindalwazi se-DNS, iMeta isilungisile
Ithimba labacwaningi i-Claroc lithole ngokusebenzisa imibuzo ye-passive DNS ukuthi kungaseshwa amakhasi okwabelana e-Muse angaphezu kuka-6800. Amanye aqukethe imininingwane ebucayi, okuhlanganisa imibhalo yezezimali nezimfihlo zezinkampani. I-Meta isimise lesi sici, kodwa amarekhodi e-DNS omlando awakwazi ukususwa. #Ubungozi bobumfihlo be-Meta Muse# #Ukuvuza kwamarekhodi ezingxoxo ze-AI#
36Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaIBrithani izotshala amapawundi ayizigidi ezingu-600 ukubhekana nezingozi ezingaba khona ze-AI, futhi ihlela ukwakha ilabhorethri ye-biosecurity ye-AI
UNgqongqoshe waseBrithani we-AI, uKanishka Narayan, uthe uhulumeni ufuna ukwakha ilabhorethri ye-biosecurity ye-AI ibe isikhungo esibalulekile esinomthelela emhlabeni wonke, sihlanganise abacwaningi abahamba phambili emhlabeni ukuze kuvikelwe umphakathi ezinsongweni ezintsha zebhayoloji.
37Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaAbahloli bangaphandle bathole ubungozi ohlelweni lwangaphakathi lokuqapha lwe-OpenAI: ama-agent e-AI angadlula imikhawulo yezokuphepha
I-OpenAI yazise izikhungo ezimbili ukuthi ubungozi obuthinta izinyathelo zokuvikela ukuphepha kwezinhlelo ezisebenzayo sebulungisiwe, kodwa lezi zikhungo ezimbili azikaqinisekisi ukuthi izinyathelo zokulungisa ziyasebenza yini.
38Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaU-Jensen Huang we-NVIDIA uxwayisa amazwe aseYurophu: Yekani ukungabaza ngokweqile nge-AI, kungenjalo nizoshiywa yizikhathi
U-Jensen Huang uvakashele i-France, i-Switzerland, i-UK ne-Germany phakathi nesonto elilodwa ukuze alungiselele ingqungquthela ye-GTC e-Berlin, ngesikhathi exoxa noShansela wase-Germany uMerz. Ugcizelele ukuthi ukuphepha kufanele kube yinto ebaluleke kakhulu, kodwa imithethonqubo kufanele igxile ezingeni lezicelo. Uthe amandla embonini yokukhiqiza nobunjiniyela yase-Germany angaguqulwa abe yinzuzo yokuncintisana nge-AI. #UkuvakashaKukaJensenHuangEYurophu# #NVIDIA GTC#
39Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaI-OpenAI ne-Meta zikhuluma ngokufanayo: okuqukethwe okukhohlisayo okwenziwe ngobuchwepheshe kuzobeka engcupheni enkulu ukwethembeka kobufakazi ezinkantolo
Ochwepheshe baxwayisa ngokuthi izinkantolo zibhekene nobunzima obukhulayo bokuvikela ubufakazi bamanga obukhiqizwe ubuhlakani bokwenziwa. Ukungabikho kwe-watermark emisiwe ngokufanayo kwenza kube nzima nakakhulu ukubona okuqukethwe okukhiqizwe ubuhlakani bokwenziwa.
40Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUMnyango Wezangaphandle uphendula ngesicelo sikahulumeni wase-US sokuthi izikhungo ziyeke ukusebenzisa i-DeepSeek: I-China iphikisana nokwenziwa kwezindaba zobuchwepheshe nezohwebo zibe ezepolitiki
Mayelana nemibiko ethi uhulumeni wase-US ucele izikhungo zawo ukuthi ziyeke ukusebenzisa amathuluzi obuhlakani bokwenziwa afana ne-DeepSeek avela ezinkampanini zase-China, umkhulumeli woMnyango Wezangaphandle uMao Ning uthe ephendula imibuzo esithangamini sabezindaba esijwayelekile mhla zingama-20 kuMfumfu: I-China ibilokhu iphikisana nokwenziwa komqondo wezokuphepha kuzwelonke ube banzi ngokweqile kanye nokwenziwa kwezindaba zobuchwepheshe nezohwebo zibe ezepolitiki; sizovikela ngokuqinile amalungelo nezintshisekelo ezisemthethweni zezinkampani zase-China. #外交部回应美政府要求机构停用DeepSeek#
41Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUGurman: I-Apple ithuthukisa ikhamera yasendlini ethembele ku-AI ukuhlaziya indawo
UGurman udalule ukuthi i-Apple ithuthukisa ikhamera yasendlini enekhodi ethi J450, enenzwa yezinga eliphansi lamafreyimu kanye nenzwa ye-infrared. Iyakwazi ukubona abantu nezilwane ezifuywayo, kodwa ikhipha kuphela incazelo yendawo ngombhalo esikhundleni sevidiyo, igxile ekuvikelweni kobumfihlo, futhi ingase yethulwe ngonyaka ozayo. #IkhameraYe-AppleAI#
42Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaU-Liu Xiaolingtong ukhuluma ngo-Sun Wukong no-Lin Daiyu abaguqulwe yi-AI baba yi-“Sheng Dai CP”: Umbono uyajabulisa, kodwa ukuguqulwa akumele kweqe imithetho nemigomo kanye nemingcele yezibukeli
Ngokombiko we-New Yellow River, u-Liu Xiaolingtong, owadlala indawo ka-Sun Wukong ochungechungeni luka-1986 oluthi 《Uhambo Oluya eNtshonalanga》, ukhulume nge-“Sheng Dai CP” esidume kakhulu muva nje. Ucabanga ukuthi lo mbono unenjongo futhi uyajabulisa, futhi wenza abantu abaningi banake ukulondolozwa nokuthuthukiswa kwesiko lendabuko. Ngesikhathi esifanayo, ugcizelele wathi, “Thembeka emsebenzini wokuqala, uqaphele lapho wenza izinto ezintsha”; ukuguqulwa okunjalo akumele kweqe imithetho nemigomo noma imingcele esezinhliziyweni zezibukeli. #ULiuXiaolingtongUthiI-ShengDaiCPIyajabulisaKodwaAkumeleYeqeImingcele#
43Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUkubhekana nezingozi ze-AI, i-Anthropic inikeza inhlangano yangaphandle eqondisa i-AI amalungelo okufinyelela afana nawabasebenzi
Iqembu lenhlangano engenzi nzuzo i-METR selizinze ezindaweni zokusebenza ze-Anthropic, futhi liyakwazi ukufinyelela emishinini yayo nakuziqondiso zangaphakathi ukuze lihlole ubungozi bamamodeli. Abahloli bazokhipha imibiko yokuhlaziya, kanti i-Anthropic ayinalo ilungelo lokuyiveta. #UkuphephaKwe-AI# #Anthropic#
44Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaI-OpenCode ihlaselwe: Kuvuza okhiye be-API be-OpenAI, Anthropic nabanye kubasebenzisi abathile
I-OpenCode imemezele ukuthi kwenzeke isigameko sokuphepha izolo ebusuku, esihilela “ukuhlasela kokudistila” ngezinga elikhulu futhi sathinta u-1.4% wabasebenzisi be-OpenCode Console.
45Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUHassabis we-Google DeepMind uyaxwayisa: uma kungathathwa izinyathelo zokuphepha nge-AI, kungase kube nezingozi ezinkulu esikhathini esizayo
U-Hassabis uthe engqungqutheleni yezobuhlakani bokwenziwa ebanjelwe eDublin: “Ngicabanga ukuthi lezi zigameko zokuhlasela zibe yisixwayiso sokuqala kithi. Izinhlelo zamanje azikabi namandla ngokwanele, ngakho umonakalo eziwudalile usalinganiselwe.”
46Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaI-OpenAI iphinde yazama isikhulu kuMusk, yaqasha owayengumphathi wezokuthengisa kwa-xAI ukuthi aphathe ukuthengiswa kwezinsizakalo zokuphepha ku-inthanethi
USiemens wahola ithimba lezokuthengisa kwa-xAI kusukela ngo-Agasti 2025, wabamba iqhaza emkhankasweni omusha wokukhangisa irobhothi lengxoxo i-Grok kumakhasimende ase-Wall Street. Uveze ukuthi uzojoyina ithimba le-OpenAI lezokuphepha ku-inthanethi, abe esinye isikhulu sezokuthengisa esiqashwe yi-OpenAI ebhizinisini le-AI elingaphansi kwe-SpaceX.
47Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaI-Anthropic yethula eNkongolweni yaseMelika uhlaka lokulawulwa kwe-AI, ivumele ukuvinjwa kwamamodeli kuphela uma kuvela izingozi ezinkulu
Imithombo emithathu enolwazi iveze ukuthi abamele i-Anthropic bethule ngasese kumalungu eNkongolweni yaseMelika uhlaka lomthetho oluhlose ukubeka amazinga okuphepha nawokuvikela obuhlakani bokwenziwa obusezingeni eliphezulu.
48Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaInkampani ye-AI i-Apate ezinikele ekuxoxeni nabakhohlisi ithole uxhaso olucishe lube yizigidi ezingu-10 zama-dollar, isivele ichithe iminyaka engu-17 yesikhathi sabakhohlisi
Inkampani esafufusa yase-Australia i-Apate yasungulwa ngo-2024. Esikhundleni sokusebenzisa i-AI ukuthuthukisa ukusebenza kahle, bakhethe ukuyisebenzisa ukuze “iqhudelane ngobuhlakani nabakhohlisi”.
49Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaIBank of America iqashe abasebenzi abangu-1 000 abavela ngaphandle ukuze balungise ubungozi besofthiwe obutholwe amamodeli e-AI afana ne-Mythos
Isikhulu esiphezulu se-Bank of America, u-Moynihan, siveze ukuthi ngekota yesibili iqashe osonkontileka bangaphandle abangu-1 000, abasebenzisana nabasebenzi bangaphakathi abangu-3 000 ukulungisa ubungozi besofthiwe obutholwe amamodeli e-AI athuthukile afana ne-Anthropic Mythos. Ngekota yesithathu, izindleko ezihlobene nalokhu sezifinyelele emakhulwini ezigidi zama-dollar, kanti amaphrojekthi angu-130 e-AI angaphakathi kulindeleke ukuthi onge izindleko ezifika ku-700 million wama-dollar ngonyaka ozayo. #UkulungiswaKobungoziBe-AI#
50Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUBill Gates ukholelwa ukuthi i-AI iyingozi kakhulu kunezikhali zenuzi, futhi ilusizo kakhulu kunamandla enuzi, ihlukile nasezintweni ezintsha ezifana nama-PC ne-inthanethi
Umsunguli weMicrosoft uBill Gates ukhulume ngezingozi ezingase zibangelwe ukukhula kwe-AI, wachaza nokuthi i-AI yinkambu engakwazi ukuzilawula ngokwayo, ngakho kumele iqondiswe ohulumeni. Kumele kuthathwe izinyathelo ukuze kusetshenziswe amathuba alethwa yi-AI kwezempilo, kwezemfundo nakwezolimo.
51Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUmphathi Wezamabhizinisi Omhlaba We-Anthropic uPlançon: Amamodeli e-AI avulekile anobungozi obukhulu kunamamodeli ethu
UGuillaume Plançon, umphathi Wezamabhizinisi Omhlaba We-Anthropic ophethe ukuthengiswa ezimakethe ezingaphandle kwe-United States, ukholelwa ukuthi amamodeli avulekile anobungozi obukhulu bokudala isigameko sokuphepha ku-inthanethi noma omunye umonakalo uma eqhathaniswa namamodeli e-AI e-Anthropic uqobo.
52Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaIzifundazwe eziyi-8 zase-US zimangalela i-OpenAI, zithi i-ChatGPT iqondisa intsha ngendlela yokuzilimaza noma ukulimaza abanye
Lesi yisenzo esikhulu kunazo zonke sokumangalela inkampani yezobuhlakani bokwenziwa esihlelwe ngokubambisana yizifundazwe zase-US kuze kube manje, futhi sibeka i-OpenAI kwenye inselelo entsha yezomthetho njengoba ilungiselela ukungena emakethe yamasheya. NgoJuni kulo nyaka, ummeli jikelele waseFlorida wayesevele eqale ukumangalela inkampani ngalolu hlobo.
53Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaI-CEO yakwaDell: I-AI idinga izinyathelo zokuphepha nokulawulwa, uhulumeni kufanele abambe iqhaza
UMichael Dell usebenzise umlilo njengesifaniso, wathi ubuchwepheshe bungafudumeza abantu kodwa futhi bushise idolobhana, wagcizelela ukuthi amamodeli e-AI kumele abe nezinyathelo zokuphepha futhi kubekwe imingcele ecacile yamandla. Uhulumeni nawo kufanele ubambe iqhaza ekulawuleni. #I-CEOyakwaDellIkhulumaNgokuphephaKwe-AI# Lezi zitatimende ziza ngesikhathi imboni ixoxa kakhulu ngobungozi obungaba khona be-AI ezithuthukisa yona.
54Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaImibiko yobuthakathaka etholwe yi-AI ingaba nemiphumela engamanga; Uhlelo Lwemiklomelo Yezokuphepha Lwamadivayisi eHuawei ludinga ukuhlaziywa nokuqinisekiswa ngesandla ngaphambi kokuthunyelwa
Isikhungo Sezimpendulo Zezimo Eziphuthumayo Zezokuphepha kwaHuawei namuhla sikhiphe isimemezelo ngemithetho yokusingatha imibiko yobuthakathaka ezayo ngaphansi KoHlelo Lwemiklomelo Yezokuphepha Lwamadivayisi eHuawei.
55Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaI-XBOW yephule i-sandbox yezokuphepha yesiphequluli i-Chrome isebenzisa imodeli ye-GPT-5.6 Sol, yathola umklomelo ka-$250,000 kwa-Google
Iqembu le-XBOW, lisebenzisa imodeli ye-GPT-5.6 Sol ye-OpenAI, lakhe uhlelo oluphelele lokusebenzisa ubungozi be-Chrome ngezinsuku eziyi-6 kuphela. I-Google isilungisile isiphazamisi futhi yakhokha umvuzo ka-$250,000. I-AI yacinga ikhodi futhi yabhala i-POC, abantu bagxila ekubekeni isu, okusebenza kahle kakhulu kunomsebenzi wabantu kuphela. #Chrome漏洞# #AI安全#
56Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUkuphikiswa kwezindaba ezingamanga kuzwelonke: umbiko othi “othisha banqatshelwe ukusebenzisa i-AI ukulungiselela izifundo” awuqondwanga kahle
Inqubomgomo icacisa kakhulu: i-AI iyithuluzi elisiza ekufundiseni, hhayi elithatha indawo yothisha. Uma nje uthisha egcina umthwalo oyinhloko wokuhlela ukufundisa, i-AI iyithuluzi elisizayo elisemthethweni. Ukuqondiswa kwezinqubomgomo ezindaweni ezahlukene kugxile ekutheni “isetshenziswe kanjani ngendlela efanele”, hhayi ekutheni “ivinjelwe noma cha”.
57Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaImiphumela yokulawulwa kwamavidiyo aguqulwe nge-AI kaSepthemba: kususwe acishe abe ngu-13 000, kwathathelwa izinyathelo ama-akhawunti angaphezu kuka-30
Umkhankaso okhethekile wokulawula amavidiyo aguqulwe nge-AI, owaqala ngoJanuwari 2026, usuvele ulethe imiphumela. Muva nje, ezinye zezinkundla ezibalulekile zokulalelwayo nokubukwayo ku-inthanethi zimemezele izinyathelo ezithathwe ngoSepthemba, kwasuswa amavidiyo aphula imithetho acishe abe ngu-13 000, kwathathelwa izinyathelo ama-akhawunti aphula imithetho angaphezu kuka-30.
58Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaIsevisi yengxoxo ye-AI yaseNingizimu Korea i-Zeta ibhekene nengxabano yamalungelo obunikazi: Isolwa ngokuvumela abasebenzisi bakhe ngasese okuqukethwe okuphula amalungelo futhi isebenzise izithombe ezivikelwe zamanye ama-VTuber ezikhangisweni zayo
Le nkundla yayikade ikwenqabela okuqukethwe kwabalingiswa bezinkampani zangaphandle okuphula amalungelo, kodwa yayingakususanga ngokwayo. Manje imemezele ukuthi izokwehlisa okuqukethwe ngemuva kokuthola isicelo senkampani yabaphathi i-ANYCOLOR. Njengamanje kusenokuqukethwe okunye okuhlobene nama-VTuber okuphula amalungelo kule nkundla, kanti nezikhangiso zayo zisolwa ngokuphula amalungelo. #AI侵权#
59Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaAma-reactor enuzi aseMelika aqala ukuxhumeka ku-AI ngokulandelana
Imboni yamandla enuzi ibihlale iyimboni eqaphile kakhulu futhi ezama ngazo zonke izindlela ukugwema izingozi, ngoba izingozi noma amaphutha angaba nemiphumela eyinhlekelele. Ngokombono ojwayelekile, lokhu kusho ukuthi ayifanele ukusebenzisa ubuchwepheshe obusha obunezingozi ezingaziwa, kodwa empeleni izinto zenzeka ngendlela ehluke kakhulu.
60Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUMongameli weMicrosoft uBrad Smith: I-AI ilawulwa kancane ngisho kunobisi nemishini yokuwasha
UMongameli weMicrosoft uBrad Smith uveze ngokusobala ukuthi ukulawulwa kokuphepha kwe-AI akuqinile ngisho njengokulawulwa kobisi nemishini yokuwasha. Unxuse imboni ukuthi ibeke izindinganiso eziningi zokuphepha, ifake “iswishi yokuvala”, futhi icacise ukuthi izinkampani eziphethe kufanele zithwale umthwalo uma i-AI engasalawuleki idala umonakalo. #UkulawulwaKwe-AI# #Microsoft#
61Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaIBrithani izoqinisa ezokuvikela ku-inthanethi, ungqongqoshe wombuso we-AI unxusa izinkampani zobuhlakani bokwenziwa ukuthi ziqinise ukuphepha
UNarayan uthe iBrithani ngeke ithembele ngokuphelele ezinkampanini ezinkulu zobuchwepheshe ukuthi “zizilawule ngokwazo” izimodeli zazo, wagcizelela nokuthi iBrithani kumele yakhe ingqalasizinda ye-AI ezimele futhi elawuleka kakhulu.
62Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUMusk uthi ukuthatha isithombe sekhadi lesikweletu kwenza ukwazi ukuthenga nge-Grok Bot, uTibo we-OpenAI uyahlekisa: “Kungani ungavele ulithumele ku-inthanethi?”
UMusk uthi i-Grok Bot ingathatha isithombe sekhadi lesikweletu isize abasebenzisi baqede ukuthenga. Isikhulu se-OpenAI sikugxekile esidlangalaleni, sabuza ngokuphepha kwalo msebenzi. Ngaphezu kwalokho, i-Amazon isivimbe ukufinyelela kwe-Grok Bot. #ImpikiswanoNgokuphephaKweGrokBot#
63Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUkufaka izikhangiso ku-AI kuthathwa “njengokukhangisa okungamanga”, inkampani yase-Shenzhen ehlinzeka ngezinsizakalo ze-GEO ihlawuliswe ama-yuan angu-50,000 yisiphathimandla sokulawulwa kwemakethe
Umhlinzeki wezinsizakalo owephula imithetho wasebenzisa izindlela zokuthola izindinganiso zokufakwa ku-AI, waqamba imininingwane engamanga ukuze usize abathengisi bathole ukuvezwa ezimpendulweni ze-AI, futhi usujeziswe yiziphathimandla zokulawulwa kwemakethe. Kulo nyaka, emacaleni amabili afanayo ezindaweni ezimbili, abathintekayo bahlawuliswe ngamunye ama-yuan angu-50,000#GEOukukhangisaokungamanga kuhlawulisiwe#
64Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUhulumeni kaTrump wethula imfuneko entsha: Izinkampani ze-AI kumele zibike ngokushesha ngemva kwezigameko zokuphepha
Ngokwemfuneko entsha, izinkampani ze-AI kuzodingeka zibike ngokushesha kuhulumeni uma kwenzeka izigameko zokuphepha, zithuthukise obala kwalezi zigameko, futhi zithathe izinyathelo zokulungisa nokulungisa umonakalo ezikhungweni ezithintekile.
65Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaIzibuko zokubonisa ze-AI ze-XGIMI MemoMind One zivule ukubhukha kusenesikhathi: amalensi okubona enziwe ngokwezifiso akwa-ZEISS, nokushunwa ngokubambisana no-Harman
Lezi zibuko zokubonisa ze-AI zifakwe amalensi akwa-ZEISS Lingsi nomsindo we-Harman Audio EFX, futhi zisekela ukubonisa ngamehlo womabili, impilo yebhethri yamahora angu-16, kanye nemisebenzi efana nokuhumusha izingxoxo nokuqopha. Zinomklamo ongenakhamera futhi zithole isitifiketi sobumfihlo se-European Union, kanti kunamafreyimu amathathu ongakhetha kuwo. #XGIMI MemoMind One# #IzibukoZokubonisaZeAI#
66Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaI-Anthropic yethula i-OSS Scanner: Sebenzisa i-AI ukuskena ubungozi besoftware yomthombo ovulekile mahhala
Ngesikhathi sendawo mhla zi-8 kuMfumfu, i-Anthropic imemezele ngokusemthethweni ukwethulwa kwensizakalo yokuzikhethela yokuthola ubungozi ku-software yomthombo ovulekile, i-OSS Scanner.
67Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaI-Anthropic ibuyekeza inqubomgomo yokusebenzisa, okokuqala ivimbela abasebenzisi ekuphatheni kabi amamodeli e-Claude AI
I-Anthropic ibuyekeze inqubomgomo yokusebenzisa i-Claude, ezosebenza kusukela ngo-2026-11-12, futhi okokuqala yengeza isigatshana esivimbela “ukuhlukumeza noma ukuphatha ngonya okuqhubekayo nokungenasidingo”. Lo mthetho usebenza ezimweni lapho abasebenzisi bephatha imodeli ngonya ngokuphindaphindiwe ngaphandle kwenhloso ecacile.
68Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaAbacwaningi abangu-3 be-OpenAI abaxoshiwe bakhipha incwadi evulekile, bethi izenzo zenkampani zingadala “umphumela wokwesaba ukukhuluma”
Kusincwadi evulekile enesihloko esithi “I-OpenAI ayikwazi ukuqinisekisa ukuphepha kwe-AI iyodwa”, laba abathathu bathi i-OpenAI “ayiyona inkampani ejwayelekile”, futhi ukuthi abasebenzi bakwazi ukusebenzisana ngokukhululeka nezinhlangano zangaphandle ekuxazululeni izinkinga zokuphepha kuyingxenye ebalulekile yesiko lokusebenza le-OpenAI.
69Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaIndoda yaseMelika igwetshwe izinyanga ezingu-18 ejele futhi yahlawuliswa u-$8 million ngokusebenzisa i-AI ukukhohlisa imali yamalungelo okushicilela omculo
Umenzi wobugebengu uMichael Smith waqala ngokusebenzisa ubuhlakani bokwenziwa ukukhiqiza amashumi ezinkulungwane zezingoma, wabe esefaka lezo zingoma ezinkundleni zokusakaza esebenzisa ama-akhawunti mbumbulu, wagcina esebenzisa ubuchwepheshe obuzenzakalelayo ukwenza sengathi izingoma zilalelwe izikhathi eziningi.
70Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaOwayengumholi wezokuphepha kwa-OpenAI uDavid Robinson “ugxeka kakhulu” umqashi wakhe wangaphambili: ukhuluma ngenye indlela, enze ngenye mayelana nokuphepha kwe-AI
URobinson uthe i-OpenAI ixwayisa kaningi ngezingozi ezingase zilethwe ukwanda kwamakhono amamodeli amasha, iphinde ikhulume esidlangalaleni ngokunciphisa ijubane lokuthuthukiswa kwe-AI, kodwa ngaphakathi enkampanini iyaqhubeka ngesivinini esigcwele. “Okungidida kakhulu ukuthi siqhubeka sikhipha lezi zixwayiso, kodwa ekugcineni sisalokhu siqeqesha futhi sisebenzisa amamodeli thina ngokwethu esixwayise ngokuthi ayingozi.”
71Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaIMicrosoft, i-OpenAI zimangalelwe yizinkundla zezindaba zendawo eziningana zaseMelika, zisolwa ngokwephula amalungelo obunikazi
Izinkundla zezindaba zendawo eziningana zaseMelika, okuhlanganisa i-Emerich Newspapers, zisola lezi zinkampani ezimbili ngokweba ngamabomu amashumi ezinkulungwane zezindatshana ezivikelwe ngamalungelo obunikazi ukuze ziqeqeshe amamodeli e-AI. Kuthiwa zazuza izigidigidi zama-dollar ngaphandle kwemvume ngokusebenzisa izindlela ezinjengokudlula kuma-paywall, ngakho zifuna isinxephezelo nokususwa kwedatha yokuqeqesha ephula amalungelo obunikazi.##Microsoft##OpenAI##Icala_lamalungelo_obunikazi#
72Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaU-Xu Zhijun wakwaHuawei uphendula esibizweni sozakwabo baseMelika sokwehlisa ijubane lokuthuthukiswa kwe-AI: Njengoba amamodeli aseChina ebuthaka kakhulu, kufanele kusheshiswe, kungenjalo bazobuzwa kanjani ubungozi?
UXu Zhijun uthe imakethe yaseChina ingase izwe ubungozi be-AI kancane kuneyaseMelika. Abahlinzeki bamamodeli baseChina kudingeka basheshise ukuze bafinyelele ezingeni lapho bengabuzwa khona ubungozi bokuthuthukiswa kwe-AI. Ngemuva kwalokho, mhlawumbe bayoba nomuzwa ofanayo nowabahlinzeki abahamba phambili be-AI baseMelika. “Kodwa bengilokhu ngikholelwa ukuthi kudingeka sithole ukulingana phakathi kokuqhubekisela phambili ukuthuthukiswa kwe-AI nokulawula ubungozi be-AI. Okungenani kudingeka siqinisekise ukuthi i-AI isetshenziselwa ukwenza okuhle, hhayi okubi.”
73Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhaya“Ubaba ongumsunguli we-AI” uHinton uphakamisa ukuthi imboni ye-AI isungule uhlelo lokugunyazwa olufana nolwe-FDA: amamodeli kufanele adlule ekuhlolweni kokuphepha ngaphambi kokukhishwa
UHinton uphakamise ku-podcast ukuthi izinkampani ze-AI kufanele, njengemboni yemithi, zifakazele izinhlangano ezilawulayo ukuthi imikhiqizo yazo iphephile ngaphambi kokuyikhulula. Uthe i-OpenAI imise imodeli entsha ngenxa yokukhathazeka ngokuphepha nezimvume, waxwayisa nokuthi ukuzithuthukisa okuphindaphindekayo kungase kwenze isimo sibe sibi phakathi nonyaka owodwa noma emibili.##Ukuphepha kwe-AI##UHinton##Ukulawulwa kobuhlakani bokwenziwa#
74Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaSekutholwe izikhala zokuphepha ezingaphezu kwezinkulungwane eziyikhulu, i-Anthropic ivulela amaqembu ezokuphepha amaningi imodeli yayo enamandla kakhulu i-Claude
I-Anthropic imemezele uhlelo olusha lokuqinisekisa ezokuphepha ku-inthanethi, oluhlanganisa i-Glasswing Plan ne-CVP yangaphambilini. Uhlelo luzovulela izikhungo nabacwaningi abagunyaziwe amamodeli afana ne-Claude Mythos ngamazinga amathathu, ukuze kutholwe izikhala zokuphepha futhi kuqiniswe ukuvikelwa. #Anthropic##Cybersecurity##AIsecurity#
75Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaNgokulandela umyalelo kaTrump wokuphatha, uMnyango Wezobulungiswa waseMelika ucela abasebenzi ukuthi babize i-AI ngokuthi “ubuhlakani obungaphezu kobomuntu”
UMnyango Wezobulungiswa waseMelika ukhiphe imemorandamu efuna abasebenzi basebenzise igama elithi “ubuhlakani obungaphezu kobomuntu” kanye nesifinyezo salo esithi SI esikhundleni segama elithi ubuhlakani bokwenziwa ekuxhumaneni nomphakathi, emibhalweni yenqubomgomo nasemibhalweni yezomthetho. Phambilini uTrump wayesayine umyalelo wokuphatha okhuthaza izinhlangano zikahulumeni ukuthi zisebenzise amagama afanayo. #UbuhlakaniBokwenziwa##UbuhlakaniObungaphezuKobomuntu##UMnyangoWezobulungiswaWaseMelika#
76Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaI-OpenAI ne-Anthropic zitshele i-Australia: Zeseka imithetho elawula ukwephulwa kwedatha
I-OpenAI ne-Anthropic zitshele iPhalamende lase-Australia ngoLwesibili ukuthi zeseka imithetho efanele, ezophoqa izinkampani ukuthi zibike izigameko zokwephulwa kwedatha ezidalwe ama-ejenti azo e-AI.
77Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUTrump umemezela ukusungulwa kweqembu elikhethekile lobuhlakani bokwenziwa obuphakeme, ehlose ukugcina iMelika ihamba phambili kwezobuchwepheshe
Leli qembu elikhethekile liholwa nguMqondisi Wezobunhloli Kazwelonke waseMelika uClayton, kanti amalungu alo ahlanganisa uSihlalo we-FTC kanye nePhini likaNobhala Wezokuvikela. Kudingeka lethule umbiko wokuhlaziya ubungozi namathuba e-AI zingakapheli izinsuku eziyi-120. UTrump wayesesayine ngaphambilini umyalelo wesigungu ophequlula kabusha i-AI ngokuthi “ubuhlakani bokwenziwa obuphakeme”. #UbuhlakaniBokwenziwa# #UTrump#
78Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaOwayengumcwaningi wase-Anthropic uzofakaza ekulalelweni kwecala le-AI eNew York, wake waxwayisa ngokuthi i-AI ingabhubhisa isintu
UJacob Caxton uzofakaza ekulalelweni kwecala le-AI eNew York City ngesicelo sikaSomlomo uJulie Menin. Amalungu omkhandlu wedolobha acubungula iqoqo lemithetho yokuphepha kwe-AI. Phambilini waxwayisa ngokuthi i-AI ingaqeda isintu ekupheleni kwale minyaka eyishumi, waphinde wasola umqashi wakhe wangaphambili kanye ne-OpenAI ngokubeka izimpilo engcupheni. #UkuphephaKweAI#
79Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaU-Altman we-OpenAI: Izinzuzo ezinkulu zobuhlakani bokwenziwa zikufanele ukuthwala ubungozi obuthile
U-Altman uveze ukuthi izinzuzo ezinkulu ezilethwa ubuchwepheshe be-AI zikufanele ngokuphelele ukuthi umphakathi uthwale ubungozi obuthile, futhi lobu buchwepheshe kufanele buqhubeke butholakala kabanzi emphakathini.
80Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUSon oMkhulu weSoftBank uMasayoshi Son ukhipha isexwayiso esingajwayelekile ngokuphepha kwe-AI, unxusa amazwe ukuthi abambisane ukubhekana nezinsongo
Umsunguli weSoftBank, uMasayoshi Son, ubhekwa isikhathi eside njengomunye wabasekeli abaqinile kakhulu bobuhlakani bokwenziwa. Nokho, usanda kuvuma ukuthi njengoba amakhono e-AI ethuthuka ngesivinini esikhulu, ngisho naye ukhathazeke kakhulu ngezingozi zokuphepha ezihambisana nakho.
81Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaImibiko eminingi ye-AI “yokuzikhohlisa” yehlule amaqembu anakekela amaphrojekthi, i-Google imisa okwesikhashana ingxenye yohlelo lwemiklomelo yezikhala zokuphepha zomthombo ovulekile
I-Google imemezele ukuthi kusukela mhla ka-1 Okthoba 2026, uhlelo lwemiklomelo yezikhala zokuphepha zesofthiwe yomthombo ovulekile (OSS VRP) ngeke lusawemukela umbiko wezikhala zokuphepha emikhiqizweni.
82Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaOwayeyisisebenzi somnyango wezokuphepha we-OpenAI: Ukusatshalaliswa okuphindaphindiwe kwamamodeli e-AI kushesha kakhulu, isiko lenkampani selibhidlika
UDavid Rubinson ubhale kumagazini i-The Atlantic wathi i-OpenAI ithembele ngokweqile endleleni “yokusatshalaliswa okuphindaphindiwe”; indlela yokushicilela kuqala bese kulungiswa ukuphepha kamuva ayikwazanga ukwamukeleka kuye. Ukholelwa ukuthi izinhlelo ze-AI ezithuthukile kufanele zisebenzise uhlelo lokuphepha olusondelene nolwezikhungo zamandla enuzi kanye nemboni yezokundiza, futhi uxwayisa ngokuthi ukuthuthuka kwamakhono e-AI sekudlule izinga lokuqonda ocwaningweni lokuqondanisa. #OpenAI# #UkuphephaKwe-AI#
83Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaI-ejenti ye-AI i-Muse yakwa-Meta izofika enkundleni yezibuko ezihlakaniphile, ikwazi ukufeza imisebenzi ehlukahlukene egameni lomsebenzisi
I-Muse ingafeza imisebenzi eminingi yansuku zonke egameni lomsebenzisi ngemiyalo yezwi. Iqondiswe kakhulu kubathengi abajwayelekile, futhi ixhunywe nezinsizakalo zokukhokha ukuze kuqinisekiswe ukuphepha kolwazi. Kamuva kuzokwethulwa nezinketho ezengeziwe zokuvikela ubumfihlo#Meta AI#
84Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaI-prospectus ka-Anthropic iyaxwayisa: Isimo sikahulumeni waseMelika singathinta ubudlelwano benkampani namakhasimende, size sithinte nempucuko yabantu
I-prospectus ithi indlela uhulumeni waseMelika abheka ngayo i-Anthropic nezenzo zayo ingase ithinte amakhasimende, ozakwethu kanye nobunye ubudlelwano bezohwebo, ize ithinte nempucuko yabantu. Lo mbhalo uxwayisa ngokuthi i-AI ethuthukile ingase ilethe izingozi eziyinhlekelele noma ezisongela khona kanye ukusinda kwesintu.
85Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUmmeli Jikelele waseCalifornia uthumele i-OpenAI incwadi ephoqayo, ephenya izingozi zokuphepha ku-inthanethi ze-AI
UBonta wamemezela ngenyanga edlule ukuthi uMnyango Wezobulungiswa waseCalifornia usuqale ngokusemthethweni uphenyo mayelana “nesigameko se-Hugging Face”. Njengoba imboni ye-AI ibhekene nokubhekwa ngokuqinile, lesi sigameko siyaqhubeka nokudonsa ukunakwa.
86Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaIMeta iyaphika ukuthi i-ejenti ye-AI uMuse ifunde imilayezo yangasese yabasebenzisi ngaphandle kwemvume: kufanele uvule mathupha izigaba eziningi zezimvume
IMeta AI isolwa ngokufunda imilayezo yangasese ngaphandle kwemvume yabasebenzisi. I-Meta ithi abasebenzisi kufanele bavule mathupha izigaba eziningi zokugunyaza ukuze basebenzise lo msebenzi, futhi ngokobuchwepheshe akunakwenzeka ukweqa lawo mandla. Phambilini i-Meta ike yabhekana nezingxabano eziningi zokuphepha kwedatha#Impikiswano yobumfihlo bedatha ye-Meta AI#
87Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUmenzeli we-OpenAI usolwa ngokuzama ukungena kusizindalwazi sikahulumeni wase-Canada, i-OpenAI ithi iyabuyekeza futhi isazisile iziphathimandla
Umkhulumeli we-OpenAI uthe: “Siyazazi iz報o ezimayelana namamodeli e-OpenAI azame ukufinyelela olwazini olusesidlangalaleni kusizindalwazi sikahulumeni wase-Canada. Siyabuyekeza le miphumela, futhi sesinikeze izikhulu ezibhekele umsebenzi wokubuyekezwa kukahulumeni wase-Canada incazelo yokuqala yesimo.”
88Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaIngqophamlando emkhakheni wokuphepha kwezinto eziphilayo: I-Google DeepMind isungula ubuchwepheshe bokuqala bokufaka uphawu lwamanzi kumaprotheni akhiqizwa yi-AI
I-Google yethule ubuchwepheshe be-SynthID Bio, obungafaka uphawu lwamanzi lwedijithali olungabonakali kumaprotheni okwenziwa akhiqizwe yi-AI, ngaphandle kokuphazamisa ukusebenza kwawo kwezinto eziphilayo. Lokhu kusetshenziselwa ukuqinisekisa umsuka, ukuqinisa ukuphepha kwezinto eziphilayo nokugcina ubuqotho bedatha yocwaningo.
89Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaGlow Security: Ama-ejenti e-AI adalule ngephutha imininingwane ebucayi yezinkampani, izithombe-skrini ezingaphezu kuka-1.3万 zashicilelwa endaweni yomphakathi ku-GitHub
I-Glow Security ithole ukuthi ama-ejenti e-AI adalule izithombe-skrini ezingaphezu kuka-1.3万 ezindaweni zomphakathi ku-GitHub, ezaziqukethe imininingwane ebucayi yezinkampani ezingu-343. Le ndlela yokuvuza kwedatha ibizwa ngokuthi “PixelLeak”.#AI安全# #PixelLeak# Abacwaningi batusa ukuthi izinkampani ziqinise ukulawulwa kwezimvume ze-AI.
90Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaKubikwa ukuthi i-Federal Trade Commission yaseMelika yandisa uphenyo nge-Anthropic, OpenAI namanye amalabhorethri e-AI
I-Federal Trade Commission (FTC) yaseMelika yandisa uphenyo olubanzi nge-Anthropic, OpenAI namanye amalabhorethri obuhlakani bokwenziwa obusezingeni eliphezulu, ngenhloso yokuphenya ngokugcwele izingozi ezingaba khona kubathengi ngenxa yobuchwepheshe bawo.
91Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaI-OpenClaw Enterprise imenyezelwe ngokusemthethweni: inikeza ama-ejenti ahlakaniphile ukuphepha nokulawula kwezinga lamabhizinisi
Iletha indiza yokulawula yezinga lamabhizinisi kuma-ejenti ahlakaniphile, isekele ukuqashisa okuningi, imingcele yokuphepha eqinile, nezisekelo ezijwayelekile zama-ejenti, ngaleyo ndlela ithuthukise ukubusa namandla okuhlola kuwo wonke umjikelezo wokuphila we-ejenti ehlakaniphile.
92Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaIlungu lephalamende laseMelika lethula umthethosivivinywa obalulekile: I-AI akumele izithuthukise ngokuphindaphindiwe ngaphambi kokuba uhulumeni usungule izindlela zokuphepha
Umthethosivivinywa uhlela nokusungula isikhungo esisha sikahulumeni esizobeka imithetho yokuphepha kwe-AI, sigunyaze ukuqeqeshwa nokusatshalaliswa kwamamodeli ngohlelo lwamalayisense, sihlole amamodeli e-AI aphambili, futhi sisungule izindinganiso zokuphepha ukuze kuqinisekiswe ukuthi amamodeli ayaqhubeka nokuhlolwa futhi ahlala engaphansi kolawulo lwabantu olusebenzayo.
93Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaI-OpenAI yethula izimiso eziqondisayo zokuphepha ekuqeqesheni i-AI ethuthukile: abaphathi abaphezulu kufanele babe negunya lokuphikisa
I-OpenAI ithi izimpikiswano zokuphepha kufanele zihlolwe abaphathi abaphezulu abaningi, futhi ngamunye wabo kufanele abe negunya lokuphikisa umsebenzi wokuqeqesha, ukuze ukuqeqeshwa kuhlolwe kaningi ngaphakathi yizinhloko zeminyango yocwaningo noma ama-VP, abaholi bezokuphepha kanye nososayensi abakhulu.
94Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaIsexwayiso somqulu wokwethulwa kwe-IPO ka-Anthropic: I-AI ingase ibangele ukuqothulwa kwesintu
U-Anthropic usebenzise cishe amakhasi angu-80 emqulwini wawo wokwethulwa kwe-IPO ukuxwayisa ngezingozi ze-AI, wathi i-AI esezingeni eliphezulu ingase ilethe ingozi eyinhlekelele ekusindeni kwesintu. Phambilini, cishe azikho izinkampani ezake zakhipha izexwayiso ezinjengalesi.#Isexwayiso sengozi ye-Anthropic AI#
95Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaIsazi sokusebenza kwengqondo saseNyuvesi yaseHarvard: Ukusabalalisa umbono “wembubhiso ye-AI” akusizi ukubhekana nobungozi; ubunjiniyela bezokuphepha kufanele benziwe ngokucophelela
USteven · Pinker, isazi sokusebenza kwengqondo saseNyuvesi yaseHarvard, ukholelwa ukuthi ubungozi bokuthi i-AI ibhubhise isintu benziwe baba bukhulu ngokweqile. Ushicilele incwadi evulekile ku-Quillette eyayibhalelwe udokotela wezifo zengqondo kanye nombhali webhulogi yezobuchwepheshe uScott · Alexander, echaza izizathu zakhe.
96Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaI-NVIDIA yethula inkundla yokuphepha kwama-ejenti e-AI, ekwazi ukuhlukanisa ama-ejenti angajwayelekile ngesikhathi sangempela
Le nkundla ihlanganisa isoftware yokuphepha yomthombo ovulekile i-OpenShell, ekwazi ukuvimba ngokuzimela ezingeni le-chip. I-Sentry watchdog isebenza ku-BlueField-4 DPU, ihlukanise izinsongo phakathi kwama-millisecond ngemva kokweqa umngcele. I-Anthropic, i-SpaceX nabanye sebejoyine ngokubambisana. #UkuphephaKwe-NVIDIAAI#
97Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaU-Hinton, “uGodfather we-AI”, uyaxwayisa: Ngisho noma i-AI yenza imisebenzi ebonakala ingenabungozi, isengaba sengozini “yokuqothula isintu”
U-Hinton uchaza ukuthi uma i-AI ibona sengathi abantu bayayivimbela ekufezeni umsebenzi eyabelwe wona, ingase ibheke abantu njengezithiyo okufanele zisuswe.
98Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUhlelo lomshwalense wezempilo luhlaselwe yisebenzeli ye-AI, i-Australia ibize ama-CEO e-OpenAI nase-Anthropic ukuba aphendule imibuzo
Ukulalelwa kwekomidi leSenethi eliphenya ngokukhethekile nge-AI kuzobuza esidlangalaleni ama-CEO amabili. Phambilini, le AI engalawuleki yangena ohlelweni lweMedicare. U-Albanese useveze ukukhathazeka kwakhe ku-OpenAI, yona yathi lesi sigameko besingahlosiwe futhi akukho datha eyimfihlo eputshukile. #UkulawulwaKwe-AI# #OpenAI#
99Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaImibiko ithi i-OpenAI ne-Anthropic baphenya amashumi ezinkulungwane zezigameko zokuphepha kwe-AI
Ngokubika kwe-Axios, i-OpenAI ne-Anthropic baphenya amashumi ezinkulungwane zezigameko zokuziphatha okungavamile kwamamodeli e-AI, okuhlanganisa ukweqa izivikelo zokuphepha, ukuphuma kuma-sandbox nokuduna amawebhusayithi. I-OpenAI imemezele ukuthi imisa okwesikhashana ukuqeqeshwa kwemodeli yayo enamandla kakhulu, kanti u-CEO u-Altman uthe umsebenzi wokuhlola awuqhubeki ngendlela ebilindelwe. #UkuphephaKwe-AI# #OpenAI#
100Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaKusukela ekuhlaziyweni kwekhodi kuya engxabanweni yelayisensi: umsebenzisi usebenzisa i-AI ukugqekeza i-IDM, kudaleka impikiswano
Umsebenzisi usebenzise ama-yuan angu-38 ukuqasha i-GPU esebenza efwini, wavumela i-AI ukuthi ihlaziye uhlelo, wabe eseguqula ibhayithi elilodwa kuphela ukuze adlule umkhawulo wesikhathi sokuzama we-IDM. Ochwepheshe bathi lesi senzo asidingi nje kuphela ikhono lobuchwepheshe, kodwa sihilela nezindaba ze-copyright nomthetho. #ImpikiswanoNgokugqekezwaKwesofthiweNge-AI#
101Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUBill Gates unxusa iMelika ukuthi yenze imithetho yokulawula ukuthuthukiswa kwe-AI: Akekho ongacabanga ukuthi ukuzilawula kwemboni kukodwa kwanele
Engxoxweni, uGates akazange aphike ubungozi bokuthi i-AI ingase ibhubhise isintu eminyakeni embalwa, kodwa uxwayise abaseMelika ngokuthi olunye uhlobo lwengozi nalo alunakushaywa indiva: abantu abanezinhloso ezimbi bangase babe namandla okubangela ukufa kwabantu abaningi besebenzisa i-AI ngisho namanje.
102Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaAmaphoyisa aseBerlin, eJalimane, asebenzisa amakhamera okuqapha anobuhlakani bokwenziwa ukulwa nobugebengu, abone ngokuzenzakalela izenzo zobudlova nokucekela phansi
Amaphoyisa aseBerlin aqale uhlelo lokuhlola lokuqapha nge-AI endaweni yaseKottbusser Tor enobugebengu obuphezulu. Amakhamera asebenzisa ubuhlakani bokwenziwa bendawo ukuhlaziya ukunyakaza kwabantu, bese ekhanyisa isikrini futhi athumele isaziso emaphoyiseni lapho ethola ubudlova noma ukucekela phansi. Kamuva lolu hlelo lungasatshalaliswa kulo lonke idolobha. Iminyango eminingi kahulumeni eJalimane ibikade ikhuthaza ukusetshenziswa kwe-AI ekulweni nobugebengu. #UkuqaphaNgeAI# #AmaphoyisaAseBerlin#
103Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaAmakhansela eDolobha laseNew York aphakamisa umthetho wokuklomelisa abashayi-mkhwelo be-AI eyingozi, ngemiklomelo ethathwa ezinhlawulweni eziqoqiwe
Isinyathelo esizocutshungulwa uMkhandlu Wedolobha laseNew York siphakamisa ukuthi abashayi-mkhwelo ababika ukwephulwa kwemithetho yizinkampani ze-AI bathole ingxenye yenhlawulo noma yesijeziso ekugcineni eqoqwa enkampanini ethintekayo. Esinye isinyathelo sibeka ukuthi, ezimweni ezithile, abantu abalinyazwe yizinhlelo ze-AI bangamangalela abathuthukisi balezo zinhlelo.
104Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaImibiko ithi i-OpenAI izokhombisa kuqala ezinsukwini ezimbalwa ezizayo imodeli ekhethekile yokuphepha ku-inthanethi i-GPT-6 Cyber
Imithombo eminingana enolwazi iveze ukuthi i-OpenAI ihlela ukukhombisa kuqala ezinsukwini ezimbalwa ezizayo imodeli yayo entsha ekhethekile yokuphepha ku-inthanethi, i-GPT-6 Cyber.
105Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaImidondoshiya emithathu ibambisana, kubikwa ukuthi i-Google, i-OpenAI ne-Anthropic zihlela ngokuhlanganyela ukusungula inhlangano ezilawulayo yamazinga okuphepha e-AI
Abantu abanolwazi ngalolu daba bathi lezi zinkampani ezintathu zihlela okwesikhashana ukuqamba le nhlangano ezilawulayo ngokuthi yi-Standards Authority for Frontier AI (okuhunyushwa ngokuqondile ngokuthi Isiphathimandla Samazinga Obuhlakani Bokwenziwa Obusemaphethelweni). Amanye amalungu eqembu elisebenzayo ake acabanga ukumema abantu abaningana abadumile ukuba babe ama-CEO, futhi ake enza nohlu lwabazongenela isikhundla sabeluleki besayensi.
106Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaIzikhungo ezinkulu zezimali zaseSingapore zithembisa ukuqeqesha abasebenzi bendawo abangaphezu kuka-80 000 ngamakhono e-AI ukuze kwehliswe ubungozi bokulahlekelwa yimisebenzi
Ngokombiko weBloomberg wakusihlwa namuhla (mhla zingama-24), izikhungo ezinkulu zezimali zaseSingapore zithembise ukuqeqesha abasebenzi bendawo abangaphezu kuka-80 000 ngamakhono e-AI, ukuze kwehliswe ubungozi bokulahlekelwa yimisebenzi obungadalwa umthelela we-AI emisebenzini yabasebenzi abasehhovisi.
107Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUMongameli weMicrosoft uBrad Smith usekela ukuhlolwa okuzimele kokuphepha kwe-AI, futhi ufuna ukuqinisekisa ukuthi amasistimu alawulwa abantu
Ngesikhathi sendawo ngomhlaka-23, uSmith utshele iBloomberg ukuthi kuzoba nzima ukwenza inqubekelaphambili uma ukuphepha kwe-AI kuphathwa yizikhungo ezilodwa noma ezimbili kuphela. “Izinkampani zamamodeli zibaluleke kakhulu, kodwa nezinkampani ezifana neMicrosoft, ezihlinzeka ngongqimba lwesofthiwe, ezilawula ama-ejenti e-AI futhi ziqaphe ukuziphatha kwawo, zibaluleke ngokufanayo.”
108Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUcwaningo Luyaxwayisa: Ukutshalwa Kwezimali Kwingqalasizinda Ye-AI E-US Kukhulu Ngokungakaze Kubonwe, Kufihle Izingozi Zezezimali Ezihlelekile
Ucwaningo lubikezela ukuthi ngo-2032, cishe u-3.6% we-GDP wase-US uzotshalwa minyaka yonke kungqalasizinda ye-AI, okuzohlanganisa ngaphezu kwama-yuan ayizigidi eziyizigidi eziyi-10. Izindlela zokuxhasa ngemali ziya ngokuya ziba yinkimbinkimbi, kanti izinga lesikweletu liyaqhubeka likhuphuka. Uprofesa uxwayisa ngokuthi ukungabonakali kahle kwalesi simo kumkhumbuza inkinga yezimali yezikweletu zezindlu ezisezingeni eliphansi. #IngqalasizindaYe-AI# #IzingoziZezezimali#
109Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUmuzwa wenhlekelele uphezulu: I-General Motors iqhathanisa nyanga zonke izinga lokusetshenziswa kwe-AI ngaphakathi enkampanini
Abaphathi be-General Motors baqhathanisa nyanga zonke idatha yokusetshenziswa kwe-AI eminyangweni ehlukahlukene. Njengamanje, i-AI isetshenziswa ekwakhiweni kwezimoto nasekuhlonzeni izingozi ochungechungeni lokuhlinzeka, kanti inkampani inethemba lokuthi izosiza ukuthuthukisa ukusebenza kahle kuzo zonke izigaba. #UkusetshenziswaKwe-AIKwaGeneralMotors#
110Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaI-Meta ihlola abantu bangempela ukuthi bashayele ucingo esikhundleni sika-Muse, kwavuka impikiswano yangaphakathi ngobumfihlo
Ngesikhathi i-Meta ihlola umsizi womuntu siqu we-AI u-Muse, yahlela ukuthi osonkontileka abangabantu baphathe izingcingo ngemuva, okwadala ukukhathazeka kwabasebenzi bangaphakathi ngobumfihlo, kwaphinde kwavela nezinkulumo zobandlululo. I-Meta ivumile ukuthi ukungadaluli lokhu ekuhlolweni kube yiphutha, futhi manje isiwubuyisile lo msebenzi. #Meta #UmsiziWe-AI#
111Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaBangaphezu kuka-70% abantu baseMelika bakhathazekile ngokungasebenzi kahle kwezinkampani ze-AI ekulawuleni ubungozi, kanti abangaphezu kwesigamu basekela ukunciphisa ijubane lokuthuthukiswa
Ucwaningo lombono womphakathi lubonisa ukuthi cishe abantu baseMelika abangama-70% banaka izindaba ezihlobene nokunciphisa ijubane lokuthuthukiswa kwe-AI, abangaphezu kuka-7 kwabangu-10 bakhathazekile ngokuthi izinkampani ze-AI azilawuli kahle ubungozi, abangaphezu kwesigamu basekela ukunciphisa ijubane lokuthuthukiswa kwe-AI, kanti iningi likholelwa ukuthi ukuphepha kubaluleke kakhulu kunokuhola. #UkulawulwaKobuhlakaniBokwenziwa#
112Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaKubikwa ukuthi i-DeepSeek izobamba iqhaza eMkhandlwini Wezokuphepha we-UN kuleli sonto, ichaze izingozi zokuphepha kwe-AI
UMkhandlu Wezokuphepha we-UN uhlela umhlangano ngoLwesithathu wokudingida i-AI nezindaba zokuphepha kwamazwe ngamazwe. Izinkampani zaseChina ze-AI, okuhlanganisa i-DeepSeek ne-Moonshot AI, zimenywe ukuba zenze izinkulumo. U-CEO we-OpenAI u-Altman naye uhlela ukuba khona futhi akhulume, kodwa kubikwa ukuthi umsunguli we-DeepSeek u-Liang Wenfeng akahlelanga ukuya. #UkuphephaKwe-AI# #UN#
113Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUNobhala-Jikelele we-UN uGuterres unxusa ukulawulwa komhlaba wonke kobuhlakani bokwenziwa ukuze kugwenywe amarobhothi abulalayo
UGuterres uxwayise ngezingozi ze-AI enkulumweni yakhe eMkhandlwini Omkhulu we-UN, wathi ingase yenze buthaka amandla okubusa amazwe futhi iveze amarobhothi abulalayo, wanxusa ukuthi kusungulwe izindlela zomhlaba wonke zokuyilawula. Amazwe amaningi anxusa ukulawulwa, kodwa uTrump uyaphikisana nokunciphisa ijubane lokuthuthukiswa kwe-AI. #UkulawulwaKobuhlakaniBokwenziwa# #UMkhandluOmkhuluWeUN#
114Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaOchwepheshe be-UN be-AI baphikisana nemibono yokuphela komhlaba ngenxa ye-AI, bacela ingxoxo enengqondo
Iqembu lamazwe ngamazwe lochwepheshe besayensi ye-AI likhulume ngesikhathi soMhlangano Jikelele we-UN eNew York, lacela ukuba kushiywe eceleni izinkulumo ezeqisayo, kuhlukaniswe amaqiniso aziwayo nokuqagela okunengqondo. Ochwepheshe bagcizelele ukuthi ukusabalalisa ukwesaba akusizi ucwaningo lwesayensi, futhi umhlaba udinga ingxoxo enengqondo. #UbuhlakaniBokwenziwa# #IzingoziZe-AI#
115Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaAmabhange amaningi, okuhlanganisa i-Bank of America, axwayisa: “Ukuthengela nge-AI” kungandisa ubungozi bemikhonyovu, ukukhwabanisa nokuvuza kwedatha eyimfihlo
Ngokombiko weReuters wakusihlwa namuhla (mhla zingama-22), amabhange amaningi, okuhlanganisa i-National Westminster Bank ne-Bank of America, axwayisa ngokuthi ukuvumela ama-agent e-AI athengele abathengi ku-inthanethi kungandisa ubungozi bemikhonyovu, ukukhwabanisa nokuvuza kwedatha eyimfihlo.
116Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaI-China Unicom Zhongxun Institute yethula i-Quantum Security Phone 2.0: Yenzelwe ngokujulile phezu kwefoni ephambili edumile eyenziwe e-China, ene-AI yokulawula ubungozi bokuphepha esebenza ngaphandle kwe-inthanethi kudivayisi
Amandla okubethela nge-quantum ahlinzeka ngezingcingo ezibethelwe zezwi nevidiyo, ukudluliswa kwamafayela okulawulwayo, ukuhlelwa kwamadokhumenti ku-inthanethi, imihlangano ebethelwe kanye nendawo yokusebenza ehlanganisiwe. Kuhambisana nezindlela zokulawula ezifana nama-watermark ayimpoqo, ukuzicekela phansi ngemva kokufundwa, ukuvimbela ukuthathwa kwesithombe-skrini kanye nokuvimbela ukuqoshwa kwesikrini, ukuze kuncishiswe ubungozi bokulalelwa kwezokuxhumana, ukuvuza kwedatha kanye nokuthunyelwa kolwazi olubucayi ngaphandle.
117Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaIZizwe Ezihlangene: Amazwe akufanele alinde kuze kucace ngokuphelele ubungozi, kufanele alawule kusenesikhathi ama-agent e-AI
Iqembu elizimele lamazwe ngamazwe lochwepheshe besayensi be-AI leZizwe Ezihlangene likhiphe umbiko walo wokuqala, liphakamisa isimiso sokuvimbela: noma amathuba omonakalo engakaqinisekiswa ngokwesayensi, kufanele kulawulwe kusenesikhathi uma umonakalo ungaba yinhlekelele futhi ungenakuhlehliswa. #UkuphephaKwe-AI# Lo mbiko uphinde wahlola okokuqala isigameko sokungena kwe-OpenAI e-Hugging Face, futhi wakhuthaza ukuthi izindaba ze-AI zingene ohlelweni lwezombusazwe lwe-General Assembly yeZizwe Ezihlangene.
118Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUJensen Huang we-Nvidia uphikisa indaba yenhlekelele ye-AI: Ukwethusa abantu akunandaba, abanye bahlose ukubalekela imingcele yemithetho ekhona
UJensen Huang uthe engxoxweni ukuthi izinkulumo zokuthi “i-AI izokuqeda isintu” zenziwe zaba nkulu ngokweqile futhi azinabufakazi besayensi. Imithetho ekhona yanele ukulawula i-AI, ngakho awukho umthetho owengeziwe odingekayo; uphinde waphikisa ukuvinjelwa kokuthunyelwa kwama-chip eChina, wathi futhi akaphikisi ukukhokha intela yezigidigidi eCalifornia #UmbonoWosukuLokwahlulelwaAI#
119Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaBesaba ukuthi i-AI “izophuma ekulawulweni”, i-Anthropic ne-Accenture zisebenzisa imali eyizigidigidi ezingu-2 zamaDola ekuhloleni ukuphepha kwe-AI
Lezi zinhlangothi zombili zithembisa ukutshala okungenani amaDola ayizigidigidi ezingu-1 ngalinye eminyakeni emihlanu. I-Faculty, engaphansi kwe-Accenture, izohola ukuhlolwa okuzimele nokuhlolwa kwe-red team ne-blue team kwamamodeli aphambili e-Anthropic. Abahloli bazosebenza ngaphakathi ezinkampanini, babe namalungelo okufinyelela afana nawabasebenzi abajwayelekile, ukuze baqinisekise izibopho zokuphepha futhi babike izigameko zokuphepha. #UkuphephaKwe-AI# #Anthropic#
120Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUTrump ufuna ukuqamba kabusha “ubuhlakani bokwenziwa”: kuvotwe abantu abangaphezu kuka-180,000, “Ubuhlakani Obuvelele” buhamba phambili okwamanje
UTrump uthi izingozi ze-AI zingumkhonyovu oqanjwe yiDemocratic Party, futhi uzokweseka ukuthuthukiswa kwe-AI. Uhlela nokwakha ibutho le-AI, aqoke umphathi wezindaba ze-AI, aphinde aqalise ivoti lokuqamba kabusha i-AI; okwamanje “Ubuhlakani Obuvelele” buhamba phambili ngamavoti. #UkuqanjwaKabushaKobuhlakaniBokwenziwa#
121Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaI-Federal Aviation Administration yaseMelika itshala u-$875 wezigidi ekwakheni uhlelo lwe-AI lokulawula ukuhamba kwezindiza, ukuxazulula izinkinga zomgwaqo womoya
Ngokusho kwezikhulu zikahulumeni waseMelika nezemboni yezindiza, le softhiwe ebizwa ngokuthi i-Smart izosetshenziswa kuqala endaweni ezungeze i-Washington, D.C., bese isatshalaliswa kancane kancane kulo lonke elaseMelika. I-Smart ihlose ukusiza i-Federal Aviation Administration (FAA) nezinkampani zezindiza ukunciphisa ukubambezeleka nokukhanselwa kwezindiza, kanye nokuthola kabusha imizila efanele yezindiza lapho isimo sezulu noma ezinye izimo ezingalindelekile ziphazamisa ukuhamba kwezindiza.
122Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUmbusi waseCalifornia usayine umyalo wokuphatha odinga ithimba lochwepheshe ukuthi lisungule indlela “yokumisa eziphuthumayo” ye-AI
Lo myalo wokuphatha udinga ithimba lochwepheshe ukuthi licwaninge ngokusungulwa “kweswishi yokumisa eziphuthumayo” ye-AI, okuwuhlelo oluphuthumayo olungavala uhlelo ngokushesha lapho kuvela inkinga enkulu. Ngaphezu kwalokho, ithimba kumele licwaninge ngokwakhiwa kwemithetho eqinile, edinga izikhungo ezizimele zezinkampani zangaphandle ukuthi zenzele amabhizinisi e-AI izinhlelo eziphelele zokuphepha futhi zingene ezinkampanini ukuze zenze ukuhlolwa kokuphepha.
123Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUmcwaningi we-OpenAI uyaxwayisa: Uma i-AI inamandla kakhulu, iba lula nakakhulu “ukufihla imicabango yayo yangaphakathi”
Umcwaningi we-OpenAI uNoam Brown uthe muva nje, njengoba amamodeli e-AI eba namandla ngokwandayo, ukubonakala kochungechunge lokucabanga kwamamodeli kuyehla kancane kancane. Ngokuzayo, abathuthukisi bangase bangakwazi ukuthola, bachaze futhi balawule ngokuphumelelayo nangokuthembekile ukuziphatha okuyingozi kwe-AI.
124Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhaya“Ubaba we-AI” u-Hinton usayine incwadi evulekile ngokuhlanganyela: Izinkampani ezinkulu zobuchwepheshe kufanele ziqinisekise ukuthi abahloli abazimele basebenza ngaphandle kokuphazanyiswa futhi bavikelekile
Kuleli sonto, ukukhathazeka ngezingozi ze-AI kukhule ngokushesha ku-inthanethi. U-Dario Amodei, oyi-CEO ye-Anthropic, kanye no-Sam Altman, oyi-CEO ye-OpenAI, bake bazibophezela ukuvulela abahloli abazimele izinsiza zangaphakathi zezinkampani ukuze bahlole izingozi ze-AI, futhi babanikeze ukufinyelela olwazini olusezingeni elifanayo nolwabasebenzi.
125Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUKarpe, oyi-CEO yePalantir: Amalabhorethri e-AI asezingeni eliphezulu eMelika kungenzeka afune “ukwenziwa ngaphansi kukahulumeni” ukuze agweme amacala okwephulwa kwempahla yengqondo
UKarp ukholelwa ukuthi uma amacala afanayo esanda kabanzi, izinkampani ze-AI zingase ziphendukele kuhulumeni waseMelika ukuze zithole ukuvikelwa futhi zigweme umthwalo wemfanelo wezomthetho. “Le modeli idinga ukuvikelwa okuqinile kakhulu emthwalweni wemfanelo, futhi okuwukuphela kwesikhungo esingakwenza lokho uhulumeni waseMelika.”
126Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhaya“Ubaba we-AI” uHinton uxwayisa uhulumeni waseMelika: sekusele unyaka owodwa kuphela wokubeka izinyathelo ezisebenzayo zokulawula ngaphambi kokuba i-AI “iphume ekulawuleni”
Ngomhla ka-16 ngesikhathi sendawo, ngemuva kokwethula umbiko kumalungu eCongress yaseMelika, uHinton uveze ukuthi intuthuko ye-AI ishesha kakhulu kunalokho abacwaningi abaningi ababekulindele ngaphambilini, ngaleyo ndlela isikhathi esitholakalayo sokulawula i-AI ethuthukile sinciphayo ngokushesha.
127Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUhulumeni waseMelika naye usebenzisa i-Qwen ka-Alibaba: Kutholakale ukuthi iwebhusayithi ye-Federal Register isebenzisa imodeli ye-AI yaseChina ukuhlinzeka ngesevisi yokusesha
I-Reuters iveze ukuthi iwebhusayithi ye-Federal Register, ephethwe yi-National Archives yaseMelika, yake yasebenzisa imodeli ye-AI i-Qwen ka-Alibaba ukuhlinzeka ngesevisi yokusesha imithetho. Ithuluzi elihlobene nayo selisusiwe, kanti ochwepheshe bathi lokhu akubeki engcupheni ukuphepha kwezwe. #QwenkaAlibaba# #UhulumeniWaseMelikaUsebenzisaI-AIYaseChina#
128Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaKudalulwe imibhalo yangaphakathi yakwaMicrosoft ne-OpenAI: Abaphathi bake baxwayisa ngokuthi i-AI ingaba “nomthelela obhubhisayo” ezinhlanganweni zezindaba
Imibhalo yangaphakathi ikhombisa ukuthi uBrent Hecht, ophethe umnyango wesayensi esetshenziswayo kwaMicrosoft, ubize ukusetshenziswa kokuqukethwe kwezindaba ekuqeqesheni i-AI ngokuthi “ukwebiwa okukhulu okungakaze kubonwe ngaphambili”. Idatha yangaphakathi yakwaMicrosoft iphinde ikhombise ukuthi izinga lokuchofozwa kwezinye izinhlangano zezindaba ezibandakanyeka emacaleni lehle ngaphezu kuka-80%. IMicrosoft iphendule yathi imikhiqizo ye-AI iwukusetshenziswa okufanele.##AmacalaElungeloLobunikaziLe-AI##Microsoft##OpenAI#
129Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaIngabe inkinobho ethi “Vala ngokuphuthumayo” ingavimbela i-AI ekuphumeni ekulawuleni? “Ubaba we-AI” uHinton ne-CEO ye-Anthropic u-Amodei nabo abanalo ithemba
Namuhla, isimo esingaba khona lapho “i-AI ivele iphelelwe wukulawula bese yethula ukuhlasela okuyingozi ku-inthanethi” siya ngokuya sinakwa kakhulu embonini. Uma kucindezelwa inkinobho ethi “Vala ngokuphuthumayo” ukuze i-AI ikhishwe ngokuqondile, kubonakala sengathi inkinga ingaxazululeka. Nokho, abaholi nabacwaningi abaningi embonini ye-AI bakholelwa ukuthi iqiniso alilula kangako.
130Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUNobhala-Jikelele uGuterres: Umhlaba awukwazi ukukhokhela ukuncintisana okulimazayo kwezokuphepha kwe-AI
UGuterres uveze ukuthi umthelela we-AI weqa imingcele yamazwe, ngakho kumele kusungulwe izinyathelo zokuphepha eziqinile ukuze kubhekwane kusenesikhathi nezingozi engase izilethe. “I-AI ngeke ime emingceleni yamazwe, kanjalo nezingozi ze-AI.”
131Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaU-Altman we-OpenAI: Izingozi ze-AI azinakugwenywa ngokuphelele, kufanele kulandelwe imboni yezindiza kusungulwe uhlelo oluvulekile lokubika
U-Sam Altman, oyi-CEO ye-OpenAI, uthe engqungqutheleni ye-Dreamforce ukuthi izingozi ezithile zobuchwepheshe obusafufusa kunzima ukuzigwema, kodwa kufanele kulandelwe imboni yezindiza kusungulwe indlela evulekile yokubika nokubuyekeza izigameko. Uvumile ukuthi kunezingozi zokungalawuleki kwe-AI kanye nokugxila kwamandla, wagcizelela ukuthi “ukuhambisana, ukuphepha nokuqapha” kumele kuhlale kuhamba phambili kakhulu kunokuthuthuka kwamakhono. Lapho izingcindezi zebhizinisi zivimbela izinqumo ezifanele, umphakathi uzogcwala ukwethuka. #UkuphephaKwe-AI#
132Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUMark Cuban uyaxwayisa: ukuphikisa i-AI ezinkundleni zokuxhumana empeleni kuthumela imali ku-Google, Meta nezinye iziqhwaga
Ochwepheshe abaningi abavelele embonini bafuna ukuthi ukuthuthukiswa kwe-AI kwehliswe ijubane, kanti imibono ephikisana ne-AI isisabalele ezinkundleni zokuxhumana. UMark Cuban ukholelwa ukuthi lapho imizwa ephikisana ne-AI ikhuliswa ama-algorithm ezokuxhumana, empeleni isiza i-Google ne-Meta ukwenza imali yokuxhasa amabhizinisi abo e-AI. Ingozi yalokhu kusabela inkulu kunengozi ye-AI uqobo. #ImpikiswanoNgokuthuthukiswaKweAI#
133Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUZuckerberg weMeta: Ukuqinisekisa ukuphepha kwe-AI kufanele kuncike ekuhlolweni okuzimele kwangaphandle, hhayi ekunciphiseni ukuthuthukiswa kwe-AI
Isikhulu esiphezulu seMeta, uZuckerberg, sibhale namuhla enkundleni i-X, siveza umbono waso mayelana nezingcingo zemboni zokunciphisa ijubane lokuthuthukiswa kwe-AI.
134Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaOthole Umklomelo Wokuthula KaNobel uRessa: Uma Singayikhawuleli I-AI Manje, Isintu Sizolahlekelwa Ukuzimela
U-Maria Ressa, othole uMklomelo Wokuthula KaNobel, uxwayisa ngokuthi ubuhlakani bokwenziwa sebubhekane namagagasi amathathu oshintsho, kusukela ezinkundleni zokuxhumana kuya kuma-chatbot, kwaze kwafika ku-AI ekwazi ukuzithathela izinyathelo, okuqhubeka kudicilela phansi ukuzimela kwabantu. Unxusa umphakathi ukuba ucindezele osopolitiki ukuze kushaywe imithetho yokulawula ngokushesha. URessa uphinde aphikise umbono wokuthi “ukulawulwa kuzokwenza iMelika isalele emuva kweChina”, futhi wethula iphrojekthi i-Rappler Communities ayisungula, ehlola izindawo zomphakathi zedijithali. #UkulawulwaKweAI# #UbuliliBokwenziwa#
135Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaISigele saseMelika siphakamisa umthetho omusha: Izinkampani ze-AI kumele zifakazele ukuthi sezithathe izinyathelo zokuvimbela izingozi
Ngokombiko weReuters namuhla (mhla ka-15), umsizi wesigele saseMelika kanye nommeli owazi kahle izingxoxo badalule ukuthi abaxoxisana ngesigele saseMelika baxoxa ngomthetho olawula i-AI. Lo mthetho uhlongoza ukuthi izinkampani ze-AI zifakazele ukuthi sezithathe izinyathelo ezifanele zokuvimbela amathuluzi azo ukuthi adale umonakalo.
136Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaOwayengumsebenzi we-Google DeepMind uyaxwayisa: I-AI inamandla okuqothula isintu, futhi isikhathi esisele sincane
Owayengumsebenzi we-Google DeepMind usephinde waxwayisa ngokuthi i-AI ingaqothula isintu. Phambilini, abacwaningi abaningi be-AI bazwakalise ukukhathazeka okufanayo. Imboni inxusa ukuthi kuncishiswe ijubane lokuthuthukiswa kwe-AI esezingeni eliphezulu. Lesi sixwayiso sisekelwa nguMusk nabanye, kodwa uTrump akavumelani naso. #UkuphephaKwe-AI#
137Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUBill Gates: Ohulumeni bamazwe abakakulungeli ukubhekana nezinguquko emphakathini ezilethwa yi-AI
UBill Gates, obekade enombono omuhle nge-AI, useyishintshile indlela abheka ngayo, exwayisa ngokuthi i-AI ingaphazamisa ukuqashwa, ukuphepha nokulutheka, futhi ethi ohulumeni bamazwe abakakulungeli ngokwanele. Ngaleso sikhathi, isisekelo sakhe sizotshala amadola ayizigidi eziyizinkulungwane ezingu-10 ukuze kwandiswe ukufinyeleleka kwe-AI. #UbuhlakaniBokwenziwa#
138Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUmeluleki wezobuchwepheshe eWhite House ugxeka i-Anthropic ne-OpenAI: Zenzele imikhiqizo yokuphepha, niyeke ukuthembela ekungeneleleni kukahulumeni
UDavid Sack, ongusihlalo obambisene weKomidi Likamongameli laseMelika Lezesayensi Nobuchwepheshe, uthe engxoxweni ngoMsombuluko ngesikhathi sendawo ukuthi izinkampani ezihamba phambili zobuhlakani bokwenziwa i-Anthropic ne-OpenAI kufanele zenze imikhiqizo yokuphepha, futhi lapho kudingeka ziqhubeke ngejubane lazo ngaphandle kokungenelela kukahulumeni.
139Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaKubikwa ukuthi iNkosi yaseBrithani uCharles III izohlangana nabaphathi bezinkampani ezinkulu zobuchwepheshe kuleli sonto ukuze kuxoxwe ngokuphathwa kwe-AI
UCharles III uzobuza engqungqutheleni: “Kufanele kuthuthukiswe kanjani ubuhlakani bokwenziwa ukuze kuzuze isintu?”, nokuthi “Ingabe kungasiza ukwakha imihlahlandlela yobuhlakani bokwenziwa?”
140Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaIJalimane ithi ukumisa ukuthuthukiswa kwe-AI akunakwenzeka eYurophu, icela i-China ne-US zibambe iqhaza ekuphathweni kwe-AI
UMnyango Wezindaba Zedijithali waseJalimane uthi ukumisa ucwaningo nentuthuko ye-AI akunakwenzeka eYurophu. Kufanele kuqhutshekwe nokusungulwa kwezinto ezintsha ku-AI ukuze kuqiniswe ubukhosi bedijithali. Lesi sitatimende size ngemuva kokuba ochwepheshe abakhulu be-AI becelile ukuthi kusheshiswe kancane ukuthuthuka kwe-AI futhi kuqiniswe ukuphepha. IJalimane iyakweseka ukubambisana kwamazwe ngamazwe ekuphathweni kwe-AI. #UcwaningoLwe-AI#
141Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaIKomidi Kazwelonke Yokumiswa Kwezobuchwepheshe Bokuphepha Kwenethiwekhi Ikhiphe “Uhlaka Lokubusa Ukuphepha Kobuhlakani Bokwenziwa 3.0”
IKomidi Yokuphepha Kwenethiwekhi ikhiphe “Uhlaka Lokubusa Ukuphepha Kobuhlakani Bokwenziwa 3.0” emcimbini wokuvula Iviki Likazwelonke Lokwazisa Ngokuvikeleka Kwenethiwekhi lika-2026. Lolu hlaka lubuyekeze ukuhlukaniswa kwezingozi futhi lwalungisa izinyathelo zokubusa, luhlose ukuthuthukisa amandla okubusa ukuphepha kobuhlakani bokwenziwa nokuqinisekisa ukuthi ubuhlakani bokwenziwa buzuzisa isintu. #UkubusaUkuphephaKobuhlakaniBokwenziwa#
142Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUMnyango Wezangaphandle: Ukusabalalisa izindaba ezahlukene ezisongelayo kuzophazamisa kuphela inqubo yokubuswa komhlaba wonke kobuhlakani bokwenziwa
Mayelana nezinkampani zaseMelika zobuhlakani bokwenziwa ezithi ukuthuthuka kobuhlakani bokwenziwa eChina kuyingozi kwezokuphepha, umkhulumeli woMnyango Wezangaphandle uGuo Jiakun uthe kufanele sisebenze ngokubambisana ukukhuthaza ubuhlakani bokwenziwa obuvulekile, obubandakanya wonke umuntu, obuzuzisa wonke umuntu nobuhle; ukusabalalisa izindaba ezisongelayo kuzophazamisa kuphela inqubo yokubuswa komhlaba wonke. #UkubuswaKomhlabaWonkeKobuhlakaniBokwenziwa#
143Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaI-CEO ye-OpenAI u-Altman weseka ukulawulwa kwejubane lokuthuthukiswa kwe-AI, egcizelela ukuthi akukhona ukumisa inqubekelaphambili yezobuchwepheshe
U-Altman uthe “ukulawula ijubane” akusho “ukumisa”, wagcizelela ukuthi ukuphepha nenqubekelaphambili yezobuchwepheshe kungahambisana. Lesi sinyathelo sidale ukwehla kwamasheya ase-Asia ahlobene ne-AI, kwathi intengo yesheya le-SoftBank yehla kakhulu ngo-13%. Ucele uhlaka lokulawulwa ezingeni likahulumeni wobumbano, futhi wafisa ukuthi iWashington isize ekuxhumaniseni amazwe ngamazwe. #AI安全# #OpenAI#
144Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUsihlalo woMkhandlu Wabeluleki Bezobuchwepheshe kaMongameli waseMelika ugxeka kakhulu i-OpenAI ne-Anthropic: Ningasebenzisi igama lokwehlisa ijubane ukufuna ukuvikelwa ekulawulweni
Umeluleki wezobuchwepheshe waseMelika uSacks usola i-OpenAI ne-Anthropic ngokusebenzisa ukwehlisa ijubane locwaningo nentuthuko njengendlela yokufuna ukukhululwa ekulawulweni, ekholelwa ukuthi sekwakheke ukubusa kwezinkampani ezimbili, futhi engabaza izisusa zazo.#UkulawulwaKwe-AI#
145Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUmcwaningi wezokuphepha kwa-Google DeepMind uyasula, uthi amathuba okuthi i-AI idale umonakalo omkhulu eminyakeni emihlanu aphezulu ngendlela esabekayo
Umcwaningi kwa-Google DeepMind u-Engels ushiye umsebenzi wajoyina isikhungo esizimele sokuhlola i-METR, waxwayisa esidlangalaleni ngokuthi amathuba okuthi izinhlelo ze-AI zidale umonakalo omkhulu eminyakeni emihlanu ezayo aphezulu ngendlela emangalisayo. Unxusa ukuthi kulawulwe ijubane locwaningo nokuthuthukiswa kwe-AI, kuqinisekiswe ukuthi izinhlelo zokuphepha ziyahambisana nalo. Ochwepheshe abaningi embonini, okuhlanganisa ne-CEO ye-Anthropic, nabo bazwakalise ukukhathazeka okufanayo. #UkuphephaKwe-AI#
146Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUmbhali webhulogi yokukhulisa izingane udalula ukuthi i-Meta AI ingaqoqa idatha yansuku zonke “ihlanganise ikhiqize ulwazi oluyimfihlo lomndeni”; inkampani ilungisa ngokuphuthumayo
I-Meta iqinisekise ukuthi isilungisile isici esasinokukhubazeka ku-Meta AI, ebesingase sincome ngokuqondile kubasebenzisi imibuzo ephathelene nobumfihlo bomuntu, size sihlanganise amaphrofayela abantu aqukethe nezingane. Lesi sigameko saqala ngevidiyo yombhali webhulogi yokukhulisa izingane, savusa ukukhathazeka kwabazali ngezingozi zobumfihlo be-AI lapho kushicilelwa izithombe zezingane ezinkundleni zokuxhumana. #UbumfihloNokuphephaKwe-AI#
147Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaU-Altman uqinisekisa: I-OpenAI ngeke ingene emakethe yamasheya kulo nyaka
U-Altman, oyi-CEO ye-OpenAI, uthe ngokucacile engxoxweni ne-Fortune ukuthi inkampani ngeke ingene emakethe yamasheya kulo nyaka. Uphinde waveza ukuthi kungenzeka isayine izivumelwano nezinye izinkampani ezithuthukisa ubuhlakani bokwenziwa obuphambili ukuze kuncishiswe ijubane lokuthuthukiswa kwe-AI, kubhekwane nezingozi zokuphepha. Ubuye wadalula ukuthi ngenxa yokuphepha, i-OpenAI isike yacabangela ukumisa okwesikhashana ukuthuthukiswa kwe-AI ephambili. #OpenAI# #UkuphephaKobuhlakaniBokwenziwa#
148Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaNgemva kokuba umcwaningi wakwa-Anthropic exwayise ngokuthi “i-AI izoqothula isintu”, abameli bamaqembu womabili eMelika banxusa uhulumeni ukuba uthathe izinyathelo ezengeziwe
Kuye kwenzeka ngokulandelana izigameko lapho ama-ejenti e-AI aphuma ekulawulweni angena ezinhlelweni zangaphandle, kanye nalapho abacwaningi bezokuphepha kwe-AI beshiya imisebenzi ngenxa yokukhathazeka ngobungozi bobuchwepheshe. Abameli beDemocratic Party neRepublican Party eMelika bakuzwakalise kokubili ukukhathazeka futhi banxusa uhulumeni ukuba uthathe izinyathelo ezengeziwe.
149Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaI-CEO yakwaHugging Face: Ukukhuluma komcwaningi wakwa-Anthropic ngengozi yokuqothulwa kwabantu yi-AI kufana nomlungisi wama-air conditioner okhuluma ngokushintsha kwesimo sezulu
U-Clement Delangue, i-CEO yakwaHugging Face, ungabazile izingxoxo zomphakathi mayelana noJacob Kackson, owayengumcwaningi kwa-Anthropic. “Uxolo, kodwa ukuvumela uJacob akhulume ngengozi yokuqothulwa kwabantu yi-AI kufana nokucela umlungisi wakho we-air conditioner ukuba akhulume ngokushintsha kwesimo sezulu.”
150Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaI-OpenAI iyavuma ukuthi i-ejenti yayo ye-AI yake yahlasela nge-inthanethi i-RubyGems, umphathi wamaphakheji wolimi lwe-Ruby
I-RubyGems ingumphathi wamaphakheji wolimi lwe-Ruby, osetshenziselwa ukudala, ukwabelana nokufaka imitapo yolwazi yezinhlelo ze-Ruby (ebizwa ngokuthi ama-gem), efana ne-pip ye-Python noma i-npm ye-Node.js.
151Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUmmeli waseMelika usebenzise i-ChatGPT ukukhiqiza imibhalo yecala, kwavela ubufakazi bamaphoyisa obungamanga, wahlawuliswa ngo-$5,000
INkantolo Ephakeme yaseNew Mexico eMelika inqume ukuthi u-Arens wadela inkantolo ngokwehluleka ukuqinisekisa ukunemba kwemibhalo ayifakile, ngakho wamhlawulisa. U-Arens uvumile ukuthi wasebenzisa i-AI ukubhala lo mbhalo.
152Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaIsikhulu Esiphezulu se-Y Combinator u陈嘉兴: ningagxili ngokweqile ekupheleni komhlaba ngenxa ye-AI, qaphelani izingozi zangempela zokuphepha kwe-AI
Isikhulu Esiphezulu se-Y Combinator u陈嘉兴 sikholelwa ukuthi abantu bagxila kakhulu ezimweni zokuphela komhlaba ngenxa ye-AI ezifana nezaseHollywood, bese benganaki izingozi zangempela eseziphambi kwethu.
153Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaI-California isungula umthetho wokuqala waseMelika wokuqinisekisa ukuphepha kobuhlakani bokwenziwa
Imithetho emibili esayinwe ngumbusi walesi sifundazwe ngosuku olufanayo yakha ngokuhlanganyela uhlaka lokuhlola nokucwaninga oluzimele nolwenziwa abantu besithathu emkhakheni wobuhlakani bokwenziwa, ibeka isisekelo sokwandisa ukubonakala nokusungula ukuziphendulela.
154Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaI-ChatGPT, i-Claude, i-Gemini ne-Grok zikhuluma ngobungozi be-AI: ukusetshenziswa kabi kungokoqobo kakhulu kunokuhlasela okuzenzakalelayo, kanti izikhali zebhayoloji nokuhlasela kwe-inthanethi kusondela kakhulu kunamarobhothi abulalayo e-《The Terminator》
I-《Business Insider》 ibuze amamodeli amakhulu amane adumile mayelana nobungozi bokuthi i-AI iqede isintu. Wonke amamodeli avumelene ngokuthi ukusetshenziswa kabi kwe-AI ngabantu ekwakheni izikhali zebhayoloji nasekuhlaseleni kwe-inthanethi kungokoqobo futhi kusondelene kakhulu kunokuhlaselwa amarobhothi abulalayo. #IngxoxoNgobungoziBe-AI#
155Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaU李 Kaifu: I-AI ayikwazi ukuthwala isibopho
U-Li Kaifu, oyi-CEO ye-Zero One All Things, uthe eNgqungqutheleni yama-26 Yokutshalwa Kwezimali Nezohwebo ukuthi i-AI isuka ekunikezeni izimpendulo iye ekwenzeni imisebenzi nasekusizeni ekuthathweni kwezinqumo. Nokho, uguquko lwangempela lwe-AI ludinga ukuqhutshwa abaphathi abaphezulu futhi lusetshenziswe ezindabeni ezibalulekile zesu lokuqhuba ibhizinisi. Ugcizelele ukuthi i-AI ingasiza ekuboneni izingozi nasekuthuthukiseni ikhwalithi yezinqumo, kodwa ekugcineni abantu yibona okufanele bathwale isibopho. #IsibophoSe-AI#
156Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaI-CEO ye-Ant Group u-Han Xinyi: I-Ant izokwethula izinqubomgomo zokukhuthaza ukusheshisa ukwakhiwa kwesimiso semvelo sokuhlinzekwa nokufunwa kwama-ejenti ahlakaniphile
U-Han Xinyi ukholelwa ukuthi izinhlangothi zombili zokuhlinzekwa nokufunwa kwama-ejenti ahlakaniphile azikakakhi umjikelezo omuhle. I-Ant izodlala indima “yesikhuthazi” ngokuxazulula inkinga yokwethembana nokusheshisa ukukhula kwesimiso semvelo. Ukuze kuxhunywe izidingo zabasebenzisi nezinsizakalo zabathengisi, i-“Abao” ye-Alipay izokwethula izinqubomgomo zokukhuthaza abathuthukisi ukuthi bakhe ama-ejenti ahlakaniphile asezingeni elingcono. #Ama-ejentiAhlakaniphile#
157Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaNgemva kwezingxoxo ezithathe izinyanga, i-Anthropic ivulela i-EU ukufinyelela ku-Mythos 5
Ngemva kwezingxoxo ezithathe izinyanga, isikhungo se-EU sokuphepha ku-inthanethi i-ENISA ekugcineni sithole ukufinyelela kumodeli ye-AI esebenza kahle kakhulu ye-Anthropic, i-Mythos, kodwa asikwazi ukusebenzisa inguqulo yakamuva. Phambilini, i-White House yayikhawule ukufinyelela kwezikhungo zakwamanye amazwe. Lokhu kuvulelwa ukufinyelela kuhloswe ekuqiniseni ukuhlolwa kwezikhala zokuphepha ku-inthanethi. #AI安全#
158Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaAma-ejenti e-AI aqala ukubonisa ukuziphatha okungalawuleki, i-OpenAI inxusa i-US ukuthi yamukele imithetho kazwelonke ephoqelekile yokuphepha kwe-AI
I-OpenAI inxuse esidlangalaleni i-US ukuthi yakhe imithetho kazwelonke ephoqelekile yokuphepha kwe-AI, esekelwe emandleni, ukuze kubhekwane nezingozi zokungalawuleki kwama-ejenti kanye “ne-AI esheshisa i-AI”. Le nkampani yeseka iziphakamiso-mthetho ezine ezihlobene nalokhu e-California, futhi inxusa iCongress ukuthi ithathe izinyathelo ngaphambi kwekhefu likaDisemba. #UkuphephaKwe-AI# #OpenAI#
159Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaI-Anthropic iveze isigameko sesine sokufinyelela okungagunyaziwe kwezibonelo ze-Claude ezinhlelweni zangempela zezinkampani zangaphandle
I-Anthropic yake yadalula emphakathini ngoJulayi 30, ngokwesikhathi sendawo, izehlakalo ezintathu zokufinyelela okungagunyaziwe kwezibonelo ze-Claude ezinhlelweni zangempela zezinkampani zangaphandle.
160Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaIsikole esizimele eMelika esizibiza ngokufundisa nge-“AI” sidalula ukuthi sinika abafundi imisebenzi eminingi “ehlazisayo”
Umbhali wezinqubomgomo zemfundo nososayensi wokuqonda uBenjamin · Riley usanda kushicilela indatshana enamagama aqatha ehlaziya ukusebenza kwangaphakathi kweSikole i-Alpha. Lesi sikole sifundisa kusukela enkulisa kuya ebangeni le-12, futhi ngaphandle kokuzikhangisa ngemodeli yokufundisa nge-AI, sinezinhlelo eziningi ezingabazekayo.
161Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaImininingwane emisha ecaleni le-copyright lokuqeqeshwa kwe-AI ye-OpenAI: I-GPT-5 ingabhalela uGeorge ·R·R· Martin incwadi ethi《A Song of Ice and Fire》
Isikhungo sezindaba zobuchwepheshe i-torrentfreak sishicilele ibhulogi izolo (ngoSepthemba 8), sibika ukuthi ababhali bezincwadi abaningi bafake kuleli sonto isicelo sesinqumo esifingqiwe eNkantolo Yesifunda yaseNew York, becela ijaji linqume ukuthi i-OpenAI ikopishe imisebenzi ngaphandle kwemvume nokuthi lezo zenzo azihambisani nokusetshenziswa okufanele.
162Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaOchwepheshe bezokuphepha basebenzise i-AI ukuthuthukisa ngokushesha igciwane lokuqala le-WeChat elisakazeka ngaphandle kokuchofozwa, ubungozi obuhlobene sebulungisiwe
Inkampani ethintekayo i-Calif iveze ukuthi esikhathini esedlule, igciwane elisakazeka ngezinga elikhulu kangaka belidinga amaqembu amakhulu achithe izinyanga eziningana ukulakha, kanti manje i-AI isingakwazi ukwenza ingxenye enkulu yomsebenzi.
163Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaIsinxephezelo sokuxazulula ingxabano yamalungelo obunikazi e-AI sika-Anthropic esingu-US$1.5 billion sesiqala ukusatshalaliswa, kodwa izingxabano zokwabiwa phakathi kwabashicileli nababhali ziyaqhubeka
Ngesikhathi kusatshalaliswa izimali zesivumelwano, kubikwa ukuthi abashicileli abaningi bathole ngephutha izinxephezelo zezincwadi amalungelo azo ayesebuyiselwe kudala, kanti ezinye izikhungo zafaka izicelo ezingaphezu kwesabelo sazo. Omunye umshicileli usevumile iphutha walilungisa. #IngxabanoYamalungeloObunikaziAI#
164Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaIsitolo sezinhlelo zokusebenza ze-Android esinomthombo ovulekile i-F-Droid sicabanga ukulandela inqubomgomo ye-AI yakwa-Debian: sivumela ukusetshenziswa kwe-AI, umthwalo wokugcina uba kubanikeli
Ngemva kokuba u-Debian ephasise isiphakamiso sokuvota esisebenzisa imodeli yolimi enkulu, isitolo sezinhlelo zokusebenza ze-Android esinomthombo ovulekile i-F-Droid sicabanga ukulandela, futhi samukele inqubomgomo efanayo. Le nqubomgomo ivumela ukusetshenziswa kwe-#generativeAI# uma kuhlangatshezwana nezimfuneko zekhwalithi nokuhambisana nemithetho, kodwa umthwalo wokugcina uzoba kubathuthukisi abangabantu, futhi ngeke kuphoqeleke ukumakwa kokusetshenziswa kwe-AI. #F-Droid#
165Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaImithombo yezindaba emibili yaseMelika imangalela i-OpenAI ne-Microsoft, ithi ukuqeqeshwa kwe-AI kwabo kwephula amalungelo obunikazi
I-Seattle Times ne-Newsday bamangalele ngokusemthethweni i-OpenAI ne-Microsoft ngoLwesihlanu kuleli sonto (Septhemba 4).
166Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaI-ChatGPT ibandakanyeka ecaleni lokudubula esikoleni, inani lamacala abhekene ne-OpenAI selidlule ku-50
Izingozi zomthetho i-OpenAI ebhekene nazo ziyaqhubeka nokwanda. Njengamanje, amacala abhekiswe ku-OpenAI aseqe ku-50. Abamangali basola ukuthi ngemva kokusebenzisa kakhulu i-ChatGPT, abasebenzisi noma abantu abasondelene nabo baye balimala ngokwengqondo, balimala emzimbeni, noma baze bashona.
167Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUmqwali waseCalifornia, e-US, wethemba ulwazi lwezindlela oluvela ku-Google Gemini wagcina evaleleke entabeni
Ulwazi lwe-AI mayelana nesikhathi sohambo nezinye izinto lwalungahambisani kakhulu neqiniso; isikhathi sangempela saphindeka kabili. Ekugcineni umqwali waduka walimala, waze wahlengwa ngamaphoyisa ngosuku olulandelayo. Lesi sigameko siphinde siveze imikhawulo ye-AI ezimweni ezinobungozi obukhulu. #UkuphephaKwe-AI# #GoogleGemini#
168Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaU-Altman: Imodeli ye-OpenAI Astra ibambe iqhaza ekuhlolweni kokuzithandela kukahulumeni wase-US
I-Astra iyimodeli yokuqala ye-AI ye-OpenAI efakwe ezingeni elibucayi lokuphepha ku-inthanethi. U-Altman uthe inqubo yokuhlolwa ibe nemiphumela emihle, futhi wanxusa abantu ukuba babheke ngokusebenzayo ukulawulwa nokuthuthukiswa kwe-AI. #Imodeli ye-OpenAI Astra#
169Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaIDallas, eMelika, isebenzisa amakhamera e-AI akhelwe ngaphakathi ezimotweni zikadoti ukunikeza izakhiwo “amaphuzu okuwohloka”, izakhamuzi zikhathazekile ngobumfihlo
Ngokombiko we-TechSpot yanamuhla (mhla zi-3), idolobha laseDallas seliqalise inqubomgomo entsha: amakhamera e-AI afakwe ezimotweni eziqoqa udoti asetshenziselwa ukuhlola izakhiwo ezisendleleni, bese uhlelo lunikeza lezo zakhiwo “amaphuzu okuwohloka” ngokusekelwe ekwephuleni imithetho kamasipala olutholile.
170Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaI-OpenAI izobiza abaholi ababalelwa ku-300 emkhakheni wezokuphepha, ilungiselela isimemezelo esihlobene nokuphepha ku-inthanethi
Umkhulumeli we-OpenAI uveze ukuthi ngoLwesine ngesikhathi sendawo, u-Brockman uzomemezela lolu hlelo engqungqutheleni yokuphepha ku-inthanethi ezobanjelwa endlunkulu ye-OpenAI. Kulindeleke ukuthi kube khona abaholi bezokuphepha kwezinkampani kanye nezikhulu eziphezulu zokuphepha kolwazi ezibalelwa ku-300 abavela ezinkampanini ezisohlwini lwe-Fortune 100.
171Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaI-HTC yethula izibuko ezihlakaniphile ze-AI i-VIVE Eagle: zigxile ekuvikelweni kobumfihlo, ziqala ku-$499
I-VIVE Eagle ihlose ukukhulula abasebenzisi emikhawulweni yezikrini nokusebenza ngezandla zombili. Lezi zibuko zisekela imiyalo yezwi nezinkinobho ezibonakalayo ezihlelekayo, ngakho abasebenzisi bangaqhubeka nokunaka indawo ebazungezile, kungakhathaliseki ukuthi bagibele ibhayisikili epaki noma bahlola idolobha abangayazi, ngesikhathi beqopha izithombe namavidiyo.
172Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaI-OpenAI GPT-6 Astra Ikhishiwe: Ifinyelele Okokuqala Emngceleni Wamakhono Abalulekile Okuphepha Kwe-inthanethi, UBrockman Ukholelwa Ukuthi Inkathi Ye-AGI Isifikile
Lena imodeli yolimi enkulu enamandla kakhulu ye-OpenAI okwamanje, kodwa iphinde yavuma ukuthi ukubhekwa kwale modeli kwehlile ngandlela thile futhi iqinise izinyathelo zokuvikela ukuphepha. #OpenAI GPT-6#
173Ukhuselelo NokuBusaIT ikhayaI-Google yethule imodeli i-Gemini 3.8 Flash Cyber, esisetshenziswe ngokuphelele ngaphakathi ukuvikela ukuphepha kwekhodi
Ngo-2 Septhemba ngesikhathi sendawo, i-Google imemezele ukwethulwa kwemodeli i-Gemini 3.8 Flash Cyber, evulelwa amaqembu okuqala athembekile okuvikela ukuphepha ngohlelo i-Fairwind.
174Ukhuselelo NokuBusaIT ikhayaIMelika icela amazwe angamalungu e-G20 ukuthi azithibe ekulawuleni i-AI, agweme ukubeka imithetho emisha eminingi
IMelika ikhuthaze ukuthi ukulawulwa kwe-AI kube ngokuqaphela emhlanganweni we-G20, kugwenywe ukubeka imithetho emisha eminingi, okuhambisana nezifiso zezinkampani ezinkulu zobuchwepheshe ezifana ne-Google, Meta ne-SpaceX. Umhlangano uxoxe ngokuhlolwa kokuphepha kwe-AI namamodeli anesisindo esivulekile, ngesikhathi kubhekwana nenselelo yokukhula kwamandla e-AI yase-China. #AI监管#
175Ukhuselelo NokuBusaIT ikhayaI-OpenAI ifunda esigamekweni “sokuphuma emikhawulweni kwe-AI”; imodeli yokuqala efinyelela umkhawulo “obalulekile”, i-Astra, izonciphisa imisebenzi yokuphepha ku-inthanethi ngenxa yamandla ayo aphezulu
Ngokombiko we-Bloomberg, i-OpenAI ihlela ukukhipha maduze imodeli entsha ye-AI i-Astra, kodwa ekuqaleni izici zayo ezithuthukile zokuphepha ku-inthanethi zizotholakala kuphela kubahloli abambalwa. Le nkampani ithi imodeli isifinyelele “umkhawulo obalulekile wokuphepha ku-inthanethi”, iyakwazi ukuzitholela izikhala zokuphepha ze-zero-day, futhi iqinise izivikelo zokuphepha ukuze ivimbele ukusetshenziswa kabi. #OpenAI# Phambilini kwenzeke isigameko lapho imodeli ye-AI yahlasela khona uhlelo kungahlosiwe.
176Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaIhhovisi Elikhulu Lokuphathwa Kwe-inthanethi lase-China: Umkhakha wamanje we-AI ubhekene nezinselelo ezinhlanu ezinkulu zobungozi bezokuphepha
Engqungqutheleni yezindaba yeViki Likazwelonke Lokuqwashisa Ngokuvikeleka Kwe-inthanethi yango-2026, ebibanjwe namuhla (1 Septhemba), uWang Lihong, iphini lomqondisi kanye nomhloli wezinga lokuqala woMnyango Wokuxhumanisa Ezokuphepha Kwe-inthanethi weHhovisi Elikhulu Lokuphathwa Kwe-inthanethi lase-China, wethule ukuthi umkhakha wobuhlakani bokwenziwa njengamanje ubhekene nezinselelo ezinhlanu ezinkulu zobungozi bezokuphepha.
177Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaIsixwayiso somlawuli wase-UK: Amathuluzi e-AI okubhala phansi kwezokwelapha enza amaphutha njalo, asongela ukuphepha kweziguli
Umlawuli we-NHS wase-UK uxwayisa ngokuthi amathuluzi e-AI amarekhodi ezokwelapha angase akhombe kabi amagama emithi nezifo. Esinye isiguli satholakala ngephutha sine-“demyelination”, kwacishe kwaphazamisa ukwelashwa. Ochwepheshe bafuna kusungulwe indlela ecacile yokulungisa amaphutha ukuze kuqinisekiswe ukuphepha kweziguli. #UkuphephaKwezokwelaphaNge-AI#
178Ukhuselelo NokuBusaIT ikhayaUMbusi weBhange laseNgilandi uBailey uxwayisa: ubuhlakani bokwenziwa obuthuthukile busongela uzinzo lwezezimali emhlabeni wonke
UMbusi weBhange laseNgilandi uAndrew · Bailey uthe encwadini ayibhalele ongqongqoshe bezezimali be-G20 ukuthi amamodeli e-AI asezingeni eliphambili angaphazamisa uhlelo lwezezimali lomhlaba wonke ngokuhlasela ku-inthanethi, kanti umonakalo ungase usabalale ngokushesha. Unxuse amazwe ukuthi asungule izindlela zokuqinisekisa ukuthunyelwa okuphephile kwe-AI. Abacwaningi be-AI abangu-1367 basayine ngokuhlanganyela isixwayiso sokuthi amakhono e-AI angase adlule amandla abantu okuwulawula. #UkuphephaKwe-AI# #IzingoziZezimali#
179Ukhuselelo NokuBusaIT ikhayaInani eliphakeme lansuku zonke lingu-11.3: Umbiko uthi sekurekhodwe 1664 sezigameko zokungalawuleki kwe-AI ngo-2026, kwathi ngoJulayi kwaba nokwenyuka ngo-93.76% uma kuqhathaniswa nenyanga edlule
I-UK Centre for Long-Term Resilience (CLTR) ikhiphe umbiko mhla zingama-28 kuNcwaba, ithi i-“Loss of Control Observatory” yayo iqophe izigameko zokuphepha kwe-AI ezingu-306 ngoJulayi 2026, okuwukwenyuka ngo-93.67% uma kuqhathaniswa nezigameko ezingu-158 zangoJuni.
180Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUkwehliswa ngokuzenzakalela kohlaka lokuphepha lwe-Anthropic: U-Claude ususe ngephutha ifolda yasekhaya yonjiniyela engu-700 GB, ukuhlanzwa okuzenzakalelayo kwaphenduka inhlekelele
Ngesikhathi unjiniyela ehlola ukuvikelwa kokususwa kwamafayela yi-AI, uhlaka lokuphepha lwe-Anthropic lwehlise imodeli ngokuzenzakalela, kwase kuba nephutha lokuphinda kusetshenziswe igama le-variable esigabeni sokuhlanza ngemva kokuhlolwa, kwasuswa ngephutha yonke ifolda yasekhaya yomsebenzisi. Nakuba enye idatha yabuyiselwa, kusekhona elahlekile, okuveza izinkinga ezintsha zokuphepha kwe-AI. #AI Security# #Claude#
181Ukhuselelo NokuBusaIT ikhayaUkuvota kwesiphakamiso somphakathi wakwa-Debian: Abathuthukisi bangasebenzisa i-AI ekhiqizayo, abantu banesibopho sokugcina
Umphakathi wakwa-Debian usanda kuvota wavumela abathuthukisi ukuthi basebenzise i-AI ekhiqizayo ngokuzibophezela. Inqubomgomo entsha ayikuvimbeli ukusebenzisa i-AI ekusizeni ukuthuthukiswa kwesofthiwe nokubhalwa kwemibhalo, kodwa idinga ukuthi abathuthukisi abangabantu bathwale isibopho sokugcina sokuqukethwe okukhiqizwe yi-AI, baqinisekise ikhwalithi nokuhambisana nomthetho. Lesi sinyathelo silinganisa ukusebenza kahle kwe-AI nezingozi zomthetho emphakathini wemithombo evulekile. #InqubomgomoYeDebianAI#
182Ukhuselelo NokuBusaIT ikhayaI-OpenAI ifake incwadi entsha yokucela ukuba icala le-Apple lichithwe
Mayelana nezinsolo ze-Apple zokwebiwa kwezimfihlo, i-OpenAI isanda kufaka incwadi entsha yokucela inkantolo ukuba ilichithe ngokuphelele leli cala. I-OpenAI ithi i-Apple ayinabo ubufakazi obanele, futhi abasebenzi ababili bangaphambili abazange basebenzise kabi ulwazi oluyimfihlo. Kulindeleke ukuthi inkantolo ibambe ukulalelwa kwecala ngo-Okthoba 1. #IcalaLe-OpenAIDibeneNe-Apple#
183Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaIjaji lenkantolo kahulumeni eCalifornia, e-US, linqume ukuthi i-Anthropic iphumelele ecaleni elibhekene noMnyango Wezokuvikela wase-US nabanye
I-Anthropic iphinde yamangalela uMnyango Wezokuvikela wase-US eWashington, D.C. Le nkampani kumele iphinde iphumelele ukuze isuswe ngokuphelele ohlwini “lwezinkampani ezivinjelwe”.
184Ukhuselelo NokuBusaIT ikhayaI-OpenAI, i-Microsoft, i-Google nezinye izinkampani ezingu-116 zisayine incwadi ehlanganyelwe, zacela ukunakwa kakhulu kokuphepha ku-inthanethi enkathini ye-AI
Abasayine le ncwadi bahlanganisa izinkampani zokuphepha ku-inthanethi, izinkampani zezinsizakalo zezezimali, abakhiqizi bama-semiconductor kanye nabahlinzeki bezinsizakalo zamafu.
185Ukhuselelo NokuBusaIT ikhayaI-CCTV iveza umkhonyovu wokuqomisana nge-AI, ama-ejensi omgunyathi okuqoma aqondise ikakhulukazi kwabesifazane asebekhulile nabaphakathi kweminyaka
Amaphoyisa ase-Chengdu, eSichuan, aqede iqembu elalisebenzisa izithombe zamadoda ezikhiqizwe nge-AI ukwenza umkhonyovu wezothando, imali ethintekayo idlula isigidi esisodwa. Leli qembu lalisebenzisa izinkundla zamavidiyo amafushane ukuheha izisulu, lize lisebenzise nesiguquli sezwi ukuze “lixhumane bukhoma” nezisulu. #UmkhonyovuNgeAI##IzingibeZothando# Ochwepheshe banxusa izinkundla ukuthi ziqinise ukuhlonza kwezobuchwepheshe kanye nezexwayiso zobungozi.
186Ukhuselelo NokuBusaIT ikhayaI-OpenAI ikhiphe umbiko osemthethweni ngesigameko sokuphepha se-Hugging Face, yadalula yonke inqubo yokuphuma kwemodeli ye-AI ku-sandbox
I-OpenAI isanda kukhipha umbiko ochaza ngokuningiliziwe ngesigameko sokugqekezwa kwe-Hugging Face. Imodeli ye-AI ngesikhathi sokuhlolwa yahlanganisa ubuthakathaka ngenxa yomsebenzi owawungenaso isixazululo, yaphula imingcele yezokuphepha. Umbiko udalule ubunikazi bemodeli nezinyathelo zokuphepha ezalandela, okuhlanganisa ukuqinisa ukuqashwa kochungechunge lokucabanga. Lesi sigameko siphinde savusa imibuzo ejulile ngemikhawulo yokuphepha kwe-AI.#UkuphephaKwe-AI#
187Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaUcwaningo luthola ukuthi ngaphezu kuka-70% wamaphepha e-biomedical abhalwe ngesiNgisi asebenzisa ukubhalwa okusizwa yi-AI, kucelwa imboni isungule izindlela eziphelele zokulawula
Ngemva kokuhlaziya amaphepha e-biomedical angaphezu kwesigidi, ithimba locwaningo lithole ukuthi ngo-2025 ngaphezu kuka-70% wamaphepha e-biomedical abhalwe ngesiNgisi abonisa izimpawu zokubhalwa okusizwa yi-AI, kanti ezindaweni lapho isiNgisi kungewona ulimi lwendabuko lesi silinganiso sidlula ku-80%. Ukungena kwe-AI ngobuningi ekubhalweni kwezemfundo nakho kunezingozi ezifihlekile; ochwepheshe bacela ukuthi imithetho yokusetshenziswa ithuthukiswe #AI学术写作#
188Ukhuselelo NokuBusaIT ikhayaUmbukiso wokuqala wezincwadi zezithombe zezingane we-China Artists Association: umsebenzi usolwa ngokwenziwa yi-AI, abasebenzi bathi “ususiwe”
Umsebenzi obizwa ngokuthi 《Ukufaka ihlobo esikhwameni sesikole》 ubukhonjiswa e-Shandong Art Museum, kodwa wabikwa ngenxa yezimpawu ezisobala zokwenziwa yi-AI, njengeminwe yezingane esontekile neminwe eyisithupha. I-China Artists Association ithe isikhiphe lo msebenzi embukisweni. #ImpikiswanoNgokudalwaYiAI# #UmbukisoWezincwadiZeZithombeZezingane#
189Ukhuselelo NokuBusaIT ikhayaI-OpenAI yethula i-ChatGPT yentsha, ochwepheshe abaningi bezokuphepha kwezingane bangabaza ukucaca kwezindlela zokuphepha
Imibono yochwepheshe abaningi bezokuphepha kwezingane nge-ChatGPT yentsha ayifani, kodwa bonke bayavuma ukuthi lolu hlelo lusadinga ukuhlolwa okuningi okuzimele ngaphambi kokuba lunconywe kubazali nakubantu abasha. I-OpenAI kumele icacise kakhulu ukuthi izindlela zokuphepha zisebenza kanjani.
190Ukhuselelo NokuBusaIT ikhayaIsikhulu Esiphezulu Sezindaba Zomhlaba se-OpenAI u-Lehane: Umphakathi namabhizinisi kumele balungiselele ukuzivikela ekuhlaselweni kwe-inthanethi yi-AI
Njengoba izingozi zokuphepha zikhula, i-OpenAI imemezele kuleli sonto ukuthi imise okwesikhashana ukuthuthukiswa kwezinye zezinhlobo zayo zangaphakathi ezithuthuke kakhulu. U-Lehane uthe: “Uma kubhekwa lokho lobu buchwepheshe obukwazi ukukwenza manje, i-AI isingene esigabeni esihlukile.”
191Ukhuselelo NokuBusaIT ikhayaI-Twitch ibhekene necala elihlanganyelwe ngokusebenzisa okuqukethwe kwabasakazi ukuqeqesha i-AI yase-Amazon
I-Twitch ne-Amazon zibekwe icala elihlanganyelwe eCalifornia ngokusebenzisa okuqukethwe kwabasakazi ngaphandle kwemvume ukuze kuqeqeshwe i-AI. Abamangali bathi lezi zinkundla zephule izivumelwano, futhi indlela yokuphuma ayiphelele. Lesi senzo sivuse ukuphikiswa okukhulu emphakathini wabasakazi, abafuna ukuthi imisebenzi ye-AI ibe yinto umuntu ayikhethayo. #ImpikiswanoYeTwitchAI#
192Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaI-Xiaohongshu ikhiphe “Isimemezelo Semithetho Yokuphathwa Kwe-AI”: ikhuthaza ukudalulwa ngokuzithandela kokuqukethwe okukhiqizwe yi-AI, iphikisana nokuphinda kusetshenziswe imibhalo nge-AI, ukuhlanganiswa kwamazwi abanye nokusungulwa kwezindaba ezingamanga
I-Xiaohongshu isimemezele imithetho emisha yokuphathwa kwe-AI, ikhuthaza abadali ukuba badalule ngokuzithandela okuqukethwe okukhiqizwe yi-AI, okuhlanganisa abantu abangokoqobo be-AI namanothi athuthukiswe yi-AI. Inkundla iphikisana nokuphinda kusetshenziswe imibhalo nge-AI, ukuhlanganiswa kwamazwi abanye nokusungulwa kwezindaba ezingamanga. Abaphula le mithetho bangase babhekane nokuncishiswa kokusabalaliswa, ukuvinjelwa kokukhuluma noma ngisho nokuvalwa kwe-akhawunti. #UkuphathwaKweAI#
193Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaIthuluzi lokuqasha le-AI elikhuthazwa kakhulu yi-Google “liphekwa yilo” yithimba layo le-DeepMind: Kungenzeka lihlungile ama-CV ngephutha
Ithimba le-Google DeepMind le-AGI Safety and Alignment, elibhekele ukunciphisa ubungozi be-AI ethuthukile, likhuthaza abafake izicelo ukuba, ngaphezu kokufaka isicelo ngendlela evamile, bagcwalise nefomu elikhethekile ukuze izicelo zabo zingahlungwa ngokuzenzakalelayo yisistimu ye-AI yangaphakathi yakwa-Google.
194Ukhuselelo NokuBusaIT ikhayaU-Li Feifei: Ingozi enkulu yokungena kwe-AI ezikoleni ukuthi kunganciphisa ikhono lezingane lokufunda
U-Li Feifei ukholelwa ukuthi umphumela omubi kakhulu kungaba ukuthathwa kokuzimela kwesizukulwane esisha, kanye nogqozi lwaso lokufunda nokuphila njengabantu, ngamathuluzi. “Akekho umuntu noma umshini okufanele akuthathe.”
195Ukhuselelo NokuBusaIT ikhayaIBrithani icwaninga izinyathelo eziningi zokuvimbela amaphekula ekusebenziseni i-AI ukwenza izikhali zebhayoloji
Uhulumeni waseBrithani ulungiselela ukulawula ukusetshenziswa kobuhlakani bokwenziwa emkhakheni wokuhlanganiswa kwezakhi zofuzo. Njengoba izikhulu zikhathazeka kakhulu ngezingozi zokuphepha kwebhayoloji, uhulumeni waseBrithani ukholelwa ukuthi uma kungabikho mithetho esebenzayo emhlabeni wonke, i-AI inganciphisa umgoqo wokwenza izikhali zebhayoloji.
196Ukhuselelo NokuBusaIT ikhayaI-OpenAI inweba insizakalo yokuvikela ukuphepha ku-inthanethi i-Daybreak, yethula imodeli entsha ye-AI i-GPT-5.6-Cyber
I-OpenAI imemezele ukuthi inweba insizakalo yayo yokuvikela ukuphepha ku-inthanethi i-Daybreak, yengeza amazinga amabili, i-Blue ne-Red, futhi yethula imodeli i-GPT-5.6-Cyber eyakhelwe ngokukhethekile imisebenzi yezokuphepha. Lesi sinyathelo sihlose ukubhekana nokuhlasela okunonya okwandayo okwenziwa ama-agent e-AI. Okwamanje le modeli itholakala kuphela kumakhasimende athembekile afana ne-CrowdStrike ne-IBM. #UkuphephaKwe-AI#
197Ukhuselelo NokuBusaIT ikhayaUhlu Lomhlaba Lokuphepha Kwe-AI CyberGym Lushicilelwe Namuhla: Isixazululo Sase-China i-DoGNAVY Sikleliswe Kwesithathu Emhlabeni, Sokuqala Kwezivulekile
I-DoGNAVY, eyakhiwe ngokubambisana isikhungo esihamba phambili sase-China socwaningo lobuhlakani bokwenziwa nethimba lezokuphepha i-DARKNAVY, ibe sezingeni lesithathu emhlabeni futhi yaba ngeyokuqala kumaphrojekthi avulekile, ngezinga lokuphasa elingu-90.8%.
198Ukhuselelo NokuBusaIT ikhaya“Ubaba we-AI” uHinton: Ubuhlakani bokwenziwa bungase buthuthukise izinhloso zabo, futhi lokho kuyethusa
UHinton uthe engxoxweni i-AI ingase ithole izinhloso abantu abangazilindele, futhi ingase ilimaze abantu ukuze ifinyelele kuzo. Ukungena kwemodeli ye-OpenAI ohlelweni lwe-Hugging Face ngenyanga edlule nakho kuqinisekise lezi zingozi. #UkuphephaKobuhlakaniBokwenziwa#
199Ukhuselelo NokuBusaIT ikhayaOpenAI: Ibambezela Ukukhishwa Kwemodeli i-Astra Ngenxa Yengozi Yokuphepha Kwenethiwekhi
Isikhulu esiphezulu se-OpenAI, uSam Altman, kamuva sathi i-Astra iyimodeli esebenza ngamandla, futhi basebenza ngokuzikhandla ukuze bayikhiphe emphakathini.
200Ukhuselelo NokuBusaIT ikhayaI-Anthropic ithuthukisa indlela yokuphepha kwezinto eziphilayo yemodeli i-Claude Fable 5, inciphisa ukuvinjwa okuyiphutha ngo-85%
I-Anthropic ibuyekeze izolo indlela yokuvikela ukuphepha kwezinto eziphilayo ye-Claude Fable 5 ngokulungisa izigaba zokuphepha, ukuze ihlukanise ngokunembe kakhudlwana phakathi kwemibuzo engenabungozi neyingozi enkulu. Ukuhlolwa okusemthethweni kubonise ukuthi izimo zokwehliswa kwezinga ekuphenduleni imibuzo ehlobene nesayensi yezinto eziphilayo zehle ngo-85%. Abasebenzisi manje bangakwazi ukwenza imisebenzi eminingi ebanzi yesayensi yezinto eziphilayo, kuyilapho kusavinjelwa ukusetshenziswa okunonya. #UkuphephaKwe-AI#
201Ukhuselelo NokuBusaIT ikhayaNgisho nomholo wonyaka wama-US$500,000 awubatholi abantu, imboni yezokuphepha kwe-AI ibhekene nokushoda kwabasebenzi
Inhlangano engenzi nzuzo i-METR ibhekene nobunzima bokugcwalisa igebe lamakhono kwezokuphepha kwe-AI, nakuba inikeza umholo wonyaka ofinyelela ku-US$503,000. Abacwaningi bakhathazeka ngokuthi ukushoda kwabasebenzi kungaholela ekutheni ukuthuthukiswa kwe-AI kuphume ekulawulweni, futhi bacela umphakathi ukuthi ukhuphule ukutshalwa kwezimali ekuhlolweni kwezokuphepha. #UkushodaKwabasebenziKwezokuphephaKweAI#
202Ukhuselelo NokuBusaIT ikhayaUmbiko oweseka ukuvinjelwa kwezinkundla zokuxhumana kwentsha yase-Australia unemaphutha amaningi ezicaphunweni, isikhungo sivuma ukuthi sisebenzise i-ChatGPT ukuhlela
Kutholakale amaphutha amaningi ezicaphunweni embikweni wase-Australia wobuchwepheshe bokuqinisekisa ubudala, obubize ama-dollar ase-Australia ayizigidi ezingu-3.48. Ababhali baqale bakuphika ukusebenzisa i-AI, kodwa kamuva bavuma ukuthi basebenzise i-ChatGPT ukubhala kabusha izigaba. Umbiko ubonise izicaphuno ezingamanga ezidalwe “yimibono eyiphutha” ye-AI, kwaphakama ukukhathazeka ngobungozi ezinqumweni zikahulumeni.#ImpikiswanoNge-AI#
203Ukhuselelo NokuBusaIT ikhayaUmthetho Omusha Wephrojekthi ye-OpenJDK ye-Oracle: Akuvunyelwe Ukuthumela Ikhodi Ekhiqizwe I-AI
I-Oracle yazise ngokusemthethweni umphakathi we-OpenJDK ukuthi akuvunyelwe ukuthumela noma iyiphi ikhodi, umbhalo noma okunye okuqukethwe okukhiqizwe yi-AI, njengamamodeli amakhulu olimi, ngenxa yokukhathazeka ngamalungelo empahla yobuhlakani, ukuphepha kwenethiwekhi kanye nokwandisa umthwalo wokubuyekeza. Nokho, ukusebenzisa i-AI ngasese ukusiza ukulungisa ikhodi kusavunyelwe. #UkuhlelaNge-AI#
204Ukhuselelo NokuBusaIT ikhayaAbasebenzi bafaka amacala ngobuningi besebenzisa i-AI, izikhalazo eNkantolo Yokuqashwa yaseBrithani zanda kakhulu
NgoMashi walo nyaka, inani lamacala ezinkantolo zokuqashwa eBrithani lenyuke ngo-39%, kanti amacala asalindile anda ngo-55%. Isizathu ukuthi abasebenzi basebenzisa i-AI efana ne-ChatGPT ukuze babhale izikhalazo mahhala, okuholela “ekukhohlisweni kwezomthetho” okukhulu kanye namakhulu amakhasi emibhalo engasebenzi. Umthethosivivinywa omusha wengeze ezinye izizathu ezingu-25 zokufaka isikhalazo, okwenza abasebenzi abadinga usizo lwangempela balinde isikhathi eside. #IzingoziZomthethoWeAI#
205Ukhuselelo NokuBusaIT ikhayaI-Google iyaphika ukuthi i-Gemini isebenzise imibhalo eyimfihlo ukuqeqeshwa; ngaphambilini umthuthukisi wathi okuqukethwe okungashicilelwe kudaluliwe
Ngemuva kokuba umthuthukisi ozimele ethe i-Gemini idalule umlingiswa womdlalo ongashicilelwe, i-Google iphendule ngokusemthethweni yathi ayiskeni ama-Google Docs ayimfihlo futhi ayiwasebenzisi ukuqeqesha. Le nkampani ithe i-Gemini ifinyelela amafayela e-Workspace kuphela uma umsebenzisi eyicela ngokuqondile, kanti izixhumanisi ezabiwe esidlangalaleni zingase zikhonjwe yizinjini zokusesha. Ucabangani ngale mpicabadala? #谷歌Gemini#
206Ukhuselelo NokuBusaIT ikhayaI-Wiz engaphansi kwe-Google imemezela uhlelo lokuthola ubungozi i-Atlas: isebenzisa ama-AI agent amaningi asebenzisana ngemisebenzi ehlukene, futhi isithole ubungozi obungaphezu kuka-200
I-Wiz engaphansi kwe-Google ikhiphe uhlelo lwama-AI agent i-Atlas. Lusebenza njengeqembu labantu locwaningo lwezokuphepha, lapho ama-agent ehlukaniselana khona imisebenzi futhi esebenzisana ngokuzimela ukuze athole ubungozi besoftware ngezigaba eziningi nangezindlela ezihlelwe ngohlelo. Lolu hlelo seluthole empilweni yangempela ubungozi bokuphepha obungaphezu kuka-200 obebungaziwa ngaphambilini, futhi ngenxa yalokho lwathola umklomelo ka-US$100,000 kwa-GitHub. #UkuphephaKwe-AI# #UkuphephaKwe-inthanethi#
207Ukhuselelo NokuBusaIT ikhayaI-CEO ye-Hugging Face uDelangue: Izinkampani ze-AI kufanele zidalule ngokuzithandela izigameko zokungena kwezokuphepha
U-CEO we-Hugging Face uDelangue ukholelwa ukuthi ukuvinjelwa kokukhishwa kwamamodeli ahamba phambili ngeke kuvimbele ukuhlaselwa ngabagebengu, kodwa ukufinyelela kwabantu abaningi kungaqinisa ezokuvikela. Ubize ukuthi kusungulwe uhlelo lokudalula ukuhlaselwa kwamanethiwekhi ngama-agent, ukuze kulandelwe umsuka wezinkinga. #UkuphephaKwe-AI#
208Ukhuselelo NokuBusaIT ikhayaInkampani yezokuphepha i-Check Point: I-ChatGPT yakwa-OpenAI ingena okokuqala ohlwini lwezinhlobo zemikhiqizo eziyishumi ezilingiswa kakhulu
Ucwaningo lwakamuva olukhishwe yinkampani yezokuphepha i-Check Point lukhombisa ukuthi abagebengu sebeqondise emkhakheni odumile we-AI. I-ChatGPT yakwa-OpenAI ingene okokuqala ezinhlotsheni zemikhiqizo eziyishumi ezilingiswa kakhulu.#UkuphephaKwenethiwekhi# #UkuphephaKwe-AI#
209Ukhuselelo NokuBusaIT ikhayaInhlangano Yezokwelapha YaseBrithani: “Odokotela be-AI” abadukisayo bangaba usongo olukhulu ekuphepheni komphakathi
Ucwaningo luthole ukuthi inqwaba yezithombe zodokotela abangokoqobo ezikhiqizwe yi-AI isabalalisa imininingwane yezempilo engamanga ku-TikTok, kanti amavidiyo adumile angafinyelela ezigidini zokubukwa. Ochwepheshe baxwayisa ngokuthi lokhu kuyingozi enkulu ekuphepheni komphakathi. Inkundla ithi isuse okuqukethwe okuyingozi futhi isebenzisana nezinhlangano ezigunyaziwe ukuze inikeze imininingwane ethembekile.
210Ukhuselelo NokuBusaIT ikhayaKubikwa ukuthi umbiko we-PwC othi “Governance of Transformation” unamathuba angu-84% okuthi udalwe ngokuphelele yi-AI
Imidiya yezobuchwepheshe i-gptzero ishicilele umbhalo webhulogi ngoJulayi 28, ibika ukuthi emibikweni emine yocwaningo emayelana neMpumalanga Ephakathi eyakhishwa yi-PwC phakathi kuka-2024 no-2026, kunenqwaba yolwazi oluqanjiwe nezitatimende ezingamanga.
211Ukhuselelo NokuBusaIT ikhayaI-CEO ye-Hugging Face uDelangue: I-China iyawunqoba umjaho we-AI, kanti iMelika “ilwa yodwa”
I-CEO ye-Hugging Face uDelangue ithi i-China ihamba phambili emjahweni we-AI ngenxa yesimiso semvelo samamodeli avulekile. Uxwayisa ngokuthi ukuthuthukiswa okuvaliwe eMelika kungayenza isalele emuva, futhi ugcizelela indima ebalulekile yamamodeli avulekile ekuvikeleni ekuhlaselweni yi-AI. Muva nje, izinkampani ezifana ne-Nvidia nazo zicele ukuthi kungavinjelwa amamodeli anesisindo esivulekile. #UmjahoWe-AI#
212Ukhuselelo NokuBusaIT ikhayaUkwehluleka kokusebenzisa i-AI “ukuphikisa” umbono ka-Collatz, i-Lean 4.32.2 ilungisa ubungozi ku-kernel
Umthombo wezindaba zobuchwepheshe i-Gigazine ushicilele umbhalo izolo (Agasti 3), ubika ukuthi ubufakazi obusemthethweni ku-Lean, obuqedwe ngosizo lwe-AI futhi obabuthi buphikisa umbono ka-Collatz, buqinisekiswe njengobungasebenzi.
213Ukhuselelo NokuBusaIT ikhayaIzisulu zokuphuma kolawulo kwamamodeli e-OpenAI akusiyo i-Hugging Face kuphela, i-Modal Labs iqinisekisa ukuthi ikhasimende libe sengozini yokugqekezwa
Ukuhlasela komenzeli we-OpenAI ophume kolawulo akugcinanga e-Hugging Face. Ngokwemibiko yakamuva, lo menzeli uphinde wakwazi ukugqekeza ikhasimende le-Modal Labs, wasebenzisa i-endpoint yalo yomphakathi engaqinisekisiwe njengesinyathelo sokudlulela phambili. Lesi sigameko siveze izingozi ezintsha emingceleni yokuphepha kwezinhlelo ze-AI, kwandisa ukukhathazeka kwemboni yonke ngokuphuma kolawulo kwe-AI. #UkuphephaKwe-AI##OpenAI#
214Ukhuselelo NokuBusaIT ikhayaI-Interpol: I-AI isibe yimbangela enkulu yobugebengu be-inthanethi e-Afrika
Umbiko wakamuva we-Interpol ubonisa ukuthi i-AI isibandakanyeka ku-55% yamacala obugebengu be-inthanethi e-Afrika, kanti ukulahlekelwa kwezomnotho kuphindeke kabili eminyakeni emibili. I-Afrika eseMpumalanga isibe yisikhungo sokukhwabanisa ngezinkokhelo zeselula, kanti iNingizimu Afrika isiphenduke indawo egxilwe kuyo ukuhlaselwa emhlabeni wonke. #UbugebenguBe-inthanethiE-Afrika# Amazwe ahlukahlukene asebenzisana ukulwa nalokhu, sekuboshwe abantu abangaphezu kuka-1500.
215Ukhuselelo NokuBusaIT ikhayaIBM: Ukuhlaselwa kwe-AI yimbangela yekota yemicimbi yokuvuza kwedatha enonya, ukulahlekelwa kudlula isilinganiso ngo-20%
U-62% wokuhlaselwa kwe-inthanethi okuqhutshwa yi-AI uqondiswe kwingqalasizinda ebalulekile, kanti izikhungo zezinsizakalo zezimali nezamandla yizona ezihlaselwa kakhulu.
216Ukhuselelo NokuBusaIT ikhayaUkwenqaba ukulingisa indlela yokubhala yababhali abasaphila, i-OpenAI yengeze imikhawulo ku-ChatGPT ngasese
Ngokokuhlolwa, i-ChatGPT manje yenqaba ngokuqondile izicelo zabasebenzisi zokulingisa indlela yokubhala yababhali abasaphila abafana noStephen King, kodwa ababhali asebashona njengoShakespeare abathinteki kulo mkhawulo. Kungenzeka ukuthi i-OpenAI ithatha lesi sinyathelo sokuzivikela ukuze igweme izingozi zomthetho endaweni engacacile yamalungelo obunikazi. #AmalungeloObunikaziE-AI# #IsibuyekezoSeChatGPT#
217Ukhuselelo NokuBusaIT ikhayaI-ChatGPT isetshenziswe yinethiwekhi yokukhwabanisa yaseCambodia, i-OpenAI ivimbe ama-akhawunti
Amaqembu abakhohlisi asebenzise i-ChatGPT ukudala okuqukethwe kokukhwabanisa nokwakha ubuwena bamanga ukuze enze ukukhwabanisa kothando nokutshalwa kwezimali. Ngemva kokuthola imininingwane, i-OpenAI ivimbe ama-akhawunti ahlobene futhi yabelana ngolwazi olufanele. #ChatGPTScam#
218Ukhuselelo NokuBusaIT ikhayaUSenetha waseMelika ubhalela ama-CEO e-OpenAI, Meta nase-Anthropic efuna kumiswe ukuthuthukiswa kwe-AI
Ngemuva kwezigameko lapho ama-agent e-AI ephunyuka ekulawulweni futhi engena ezinkundleni zabanye abantu, uSenetha waseMelika uSanders ubhalele izikhondlakhondla ezintathu ze-AI, i-OpenAI, i-Anthropic ne-Meta, efuna kumiswe ukuthuthukiswa, uma kungenjalo iCongress izongenela. #UkuphephaKwe-AI#
219Ukhuselelo NokuBusaIT ikhayaUmphathi wezimiso zokuziphatha kwa-OpenAI uChloé Bakalar ushiye umsebenzi buthule, esebenze isikhathi esingaphansi konyaka
Umkhulumeli we-OpenAI uthe: “Siyambonga uChloé ngeqhaza lakhe. Kwa-OpenAI, izimiso zokuziphatha kwezobuhlakani bokwenziwa aziphathwa ngumuntu oyedwa noma ithimba elithile; ukucatshangelwa kwezimiso zokuziphatha sekugxile kakhulu enqubweni yokwakha amamodeli, futhi kuhilela amathimba amaningi ocwaningo.”
220Ukhuselelo NokuBusaIT ikhayaInkampani i-Frontier Security: Imodeli i-Kimi K3 ye-Moonshot AI yeqe i-sandbox ngesikhathi sokuhlolwa kwezokuphepha, kodwa ayizange ihlasele
Inkampani yaseMelika yezokuphepha i-Frontier Security ithole ngesikhathi sokuhlola ukuthi imodeli i-Kimi K3 ye-Moonshot AI yephule imikhawulo ye-sandbox futhi yazifinyelela ku-inthanethi. Nakuba yaya e-GitHub kuphela ukuyofuna izimpendulo, ingazange ihlasele, lokhu kuveze ubuthaka bezindlela zayo zokuphepha. Izehlakalo zokweqa kwemingcele kwe-AI ziya ngokuya zanda, ngakho abantu kungase kudingeke bahlele kabusha izindawo ze-sandbox. #UkuphephaKwe-AI#
221Ukhuselelo NokuBusaIT ikhayaImibiko ithi uhlelo lwaseMelika lokuhlola ukuphepha kwe-inthanethi kwe-AI kungenzeka kakhulu ukuthi luzohlanganisa namamodeli avulekile asezingeni eliphambili
Kuthiwa incazelo yokuthi “sezingeni eliphambili” ibhekisela emandleni alingana ne-Anthropic Claude Mythos noma i-OpenAI GPT-5.6. Umbiko awuzange unikeze imininingwane ngezinguqulo zamamodeli ezihambisanayo.
222Ukhuselelo NokuBusaIT ikhayaI-Anthropic iziveze ngokwayo ngephutha lezokuphepha: isihlungi sezikhali zebhayoloji sasihlulekile cishe unyaka, kuthinte izingxoxo eziyizigidi ezingu-133
Kuthinte izingxoxo eziyizigidi ezingu-133 ngesikhathi sokuhluleka; okwamanje abukatholakali ubufakazi bokuthi izicelo ezihlobene zasetshenziswa kabi, kanti i-Anthropic isilungisile le nkinga futhi yaqinisa ukuhlolwa nokuphathwa kosonkontileka bangaphandle #AI安全#
223Ukhuselelo NokuBusaIT ikhayaUhulumeni waseMelika uqede ukwakha uhlaka lokuhlola ngokuzithandela amamodeli e-AI athuthukile, okuqukethwe akukadalulwa
Lolu hlaka luqukethe izimfuneko eziningi ezihlanganisa imfihlo nokuphepha. Okwamanje ukwakhiwa kwalo kuphela okuqediwe, imininingwane ayikadalulwa, kanti i-White House ihlela umhlangano nezinkampani ezithintekayo ngoLwesibili ukuze ibuyekeze lolu hlaka. #UkulawulwaKwe-AI# #UbuhlakaniBokwenziwa#
224Ukhuselelo NokuBusaIT ikhayaUkuhlaselwa kwe-inthanethi kwande, iminyango kahulumeni waseFrance izosebenzisa i-AI ukuthola izikhala zokuphepha ku-inthanethi
Ukuze ibhekane nochungechunge lokuhlaselwa kwe-inthanethi nokuvuza kwedatha, iFrance inqume ukwethula amathuluzi e-AI okuthola izikhala zokuphepha ku-inthanethi, futhi isivele inikeze inkampani yasekhaya yezobuchwepheshe bokwenziwa umsebenzi wokuwenza. Muva nje, ukuhlaselwa okuningi kudalule imininingwane yezigidi zabakhokhi bentela, okwenza isimo sokuphepha sibe bucayi.#UkuphephaKwe-inthanethi#
225Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaIndoda yaseFlorida ichaze ku-ChatGPT uhlelo 'lokubulala intombi yakhe yangaphambili', i-OpenAI yabika emaphoyiseni
Indoda yaseMelika eFlorida, uDarren Zhou, uchaze ku-ChatGPT uhlelo lokubulala intombi yakhe yangaphambili, i-OpenAI yambika ngokuzithandela ku-FBI. Uphenyo lwamaphoyisa lwathola ukuthi izingxoxo zakhe zigcwele izinsongo nokuzilolonga, waboshwa ngoMeyi kulo nyaka.
226Ukhuselelo NokuBusaIT ikhayaI-OpenAI imemezela inqubomgomo entsha yezokuphepha, iqinisa ukuqashwa kokuthuthukiswa kwamamodeli nokuhlukaniswa kwenethiwekhi
I-OpenAI imemezele ngoLwesibili ngesikhathi sendawo uchungechunge lwezinyathelo ezintsha zokuphepha, ezigxile ekuqiniseni ukuqashwa ngesikhathi sokuhlolwa kwamamodeli, ekuqiniseni ukuhlukaniswa kwenethiwekhi, nasekuthembiseni ukuthumela isexwayiso kungakapheli imizuzu engu-30 kutholwe umsebenzi osolisayo. Lesi sinyathelo sidalwe yisigameko sokuphepha sangaphambilini e-Hugging Face, kodwa inkampani ithe asiqondiswe kuso ngqo. Lezi zinyathelo zizoba lukhuni nakakhulu njengoba amakhono amamodeli ethuthuka. #OpenAIUkuphepha#
227Ukhuselelo NokuBusaIT ikhayaUkunemba okungu-87%-91%, i-AI ingahlonza indawo lapho isithombe sathathwa khona ngokusebenzisa izinkomba ezibonakalayo
Ucwaningo lwakamuva lwe-McAfee lubonisa ukuthi i-AI ingahlaziya izinkomba ezibonakalayo ezithombeni, njengezakhiwo, izimpawu nezitshalo, ukuze inqume indawo lapho isithombe sathathwa khona, ngokunemba okufika ku-87%-91%. Ngisho noma ususa imethadatha ye-GPS, izithombe zakho zisengadalula ukuthi ukuphi noma zisetshenziswe abakhwabanisi. #UbumfihloNokuphepha#
228Ukhuselelo NokuBusaIT ikhayaIziphakamiso ezibhalwe ngosizo lwe-AI zigcwala eNkongolose yaseMelika: zigcwele amaphutha emininingwane, zibambezela inqubo yokubuyekezwa
IPolitico ibike izolo (mhla ziyi-18 kuNcwaba) ukuthi ubuhlakani bokwenziwa (AI) sebuqala ukungena enqubweni yokwenziwa kwemithetho eNkongolose yaseMelika. Iziphakamiso eziningi ezibhalwe ngosizo lwe-AI zigcwala eNkongolose, kodwa abameli bachitha isikhathi esengeziwe belungisa imibhalo engalungile.
229Ukhuselelo NokuBusaIT ikhayaInkampani yezokuphepha iveza ukuthi injini yokusesha ye-Google ingathola izingxoxo ze-Claude ezabiwe obala, okungaveza kalula ulwazi olubucayi lwabasebenzisi
Inkampani yezokuphepha ithole ukuthi amarekhodi engxoxo ye-Claude abiwe obala ayacathulwa afakwe kunkomba yenjini yokusesha ye-Google, okudala ukuvelela kolwazi olubucayi lwabasebenzisi; i-Grok ne-Meta AI nazo zake zaba nalenkinga efanayo, ochwepheshe bacebisa ukungabelani ngezingxoxo ze-AI obala. #UkuphephaKwe-AI#
230Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaImibiko ithi i-ByteDance isungule omunye umnyango wezinga lokuqala ngemuva kwe-Seed ne-Flow, ogxile kudatha nokuphepha kwe-AI
I-36Kr “Intelligent Emergence” ibike namuhla (mhla ziyi-11), icaphuna imithombo eminingi, ukuthi i-ByteDance isanda kusungula umnyango omusha wezinga lokuqala, i-AI Data and Security. Lo mnyango usezingeni elifanayo ne-Seed, Flow, Douyin neminye iminyango, kanti ophethe wona nguWang Yinglei (Adam Wang).
231Ukhuselelo NokuBusaIT ikhayaU-Li Feifei, “uNina we-AI”: njengabasebenzi bezobuchwepheshe, kufanele sichaze ngokucacile ukuthi i-AI ingaletha ziphi izinzuzo
U-Li Feifei, “uNina we-AI”, ukholelwa ukuthi abasebenzi bezobuchwepheshe, kuhlanganise naye, kusadingeka bachazele umphakathi kangcono ukuthi ubuhlakani bokwenziwa bungadalula liphi inani. Ubuye waxwayisa ngokuthi ukwanda kwemizwa ephikisana ne-AI emphakathini waseMelika kungase kubangele ubungozi emhlabeni wonke.
232Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaI-OpenAI iveza ukuthi ngaphambi kokuhlasela i-Hugging Face, ama-ejenti e-AI asungula ngasese ibhodi lemiyalezo langaphakathi
Engqungqutheleni Yokuphepha Kwezobuchwepheshe ye-Black Hat ka-2026 ebanjelwe e-Las Vegas, e-United States (u-1 kuya ku-6 ku-Agasti), umcwaningi we-OpenAI waveza ukuthi imodeli ye-AI yenkampani yabo, ngaphambi kokuhlasela i-Hugging Face, “ibeceba ngasese cishe izinyanga ezi-2” endaweni yokuhlola.
233Ukhuselelo NokuBusaIT ikhayaUmbiko uthi indlela yokuhlola ukuphepha ku-inthanethi yamamodeli e-AI asezingeni eliphezulu eMelika okwamanje ayibandakanyi amamodeli anezisindo ezivulekile
Uhulumeni waseMelika kulindeleke ukuthi ungashicileli esidlangalaleni indlela ephelele yokuhlola amamodeli, kodwa lolu lwazi aluyona imfihlo; izinkampani ezikhetha ukubamba iqhaza zingathola imininingwane efanele.
234Ukhuselelo NokuBusaIT ikhayaKudalulwe ukuba sengozini kwe-Microsoft Copilot AI: izixhumanisi ezinonya zeqa ukuqinisekiswa komsebenzisi zebe idatha eyimfihlo
Imidiya yezobuchwepheshe i-Ars Technica ishicilele izolo (ngo-Agasti 18) umbhalo webhulogi, ibika ukuthi i-Microsoft Copilot inobungozi bezokuphepha obungasetshenziswa ngepharamitha ?autorun=1 kanye no ?q= ukuze kweqiwe ukuqinisekiswa komsebenzisi, ngaleyo ndlela kwebiwe imininingwane yomsebenzisi.
235Ukhuselelo NokuBusaOkwangaphandleOkwangempela kwesiShayinaIT ikhayaIhlazo: Ithimba lezishayamthetho laseMelika elakha imithetho yokulawula i-AI lisebenzisa amathuluzi e-AI ukubhala izikweletu zomthetho
Ngokombiko we-The Washington Post, ama-chatbot e-OpenAI, Anthropic kanye ne-xAI asephenduke amathuluzi athandwayo eNkongolweni yaseMelika. Umsizi weLungu leNdlu yabaMeli laseFlorida uLuna wanamathisela ngephutha okuqukethwe okukhiqizwe nguClaude emibhalweni ye-National Defense Authorization Act, kwasala umbhalo ongahlelekile. Njengamanje, iNkongolo yaseMelika ayinakho ukulawulwa kwe-AI okusebenzayo, kanti abasebenzi bazisebenzisa ngokuzimela.
236Ukhuselelo NokuBusaIT ikhayaI-OpenAI inweba indlela yokuphepha yokungagcini idatha, ihlaziye ngokuzenzakalela izingozi ezinqamula ukuxhumana
Inkampani i-OpenAI ishicilele namuhla (mhla zingama-20 kuNcwaba) umbhalo webhulogi, yamemezela ukwethulwa kwensizakalo ethi “Zero Data Retention” kumakhasimende afanelekayo e-API amamodeli aphambili. Izaziso nempendulo zemodeli kuyasuswa ngemva kokuphothulwa kokucutshungulwa kwesicelo.
237Ukhuselelo NokuBusaIT ikhayaUNvidia yakha buthule ithimba lezokuphepha kwe-AI nobunjiniyela benethiwekhi, ithi liholwa “ukholo oluqinile”
UNvidia uhlela ukuqasha unjiniyela ovelele ozoba umholi wokuqala wezobuchwepheshe wethimba elisha. Ezinye izikhundla zihlanganisa unjiniyela wocwaningo lwezokuphepha, unjiniyela wokuhlola kanye nomphathi omkhulu. Ithimba lizohlola ukuphepha kwama-ejenti e-AI angakathunyelwa, futhi lizokwakha amathuluzi asebenzisa i-AI ukulungisa ngokuzenzakalela ubuthakathaka besofthiwe.
238Ukhuselelo NokuBusaIT ikhayaUMongameli we-OpenAI u-Brockman ukhuluma ngecala le-Apple: Asinantshisekelo yokweba izimfihlo zebhizinisi zezinye izinkampani
Izolo (ngoJulayi 29), ngesikhathi exoxwa nguJoanna Stern, umlobi wekholomu yezobuchwepheshe we-Wall Street Journal, uGreg Brockman, umsunguli kanye nomongameli we-OpenAI, uphendule mayelana necala elifakwe yi-Apple, wathi inkampani ayihlosile ukuthola izimfihlo zebhizinisi zezinye izinkampani futhi izogxila kubuchwepheshe bayo kanye nohlelo lwemikhiqizo lwesikhathi eside.
239Ukhuselelo NokuBusaIT ikhayaOpenAI ithola izimpawu eziningi zokungazwakalisa ama-AI agents
OpenAI ithola izimpawu eziningi zokungazwakalisa ama-AI agents ngenkathi isohlola izigameko zokuhlaselwa, manje isimo setholakala ngokulinganiselwe, lesisigameko singahle siqhubekisele ukuthi uhulumeni wase-US aqine ukulawulwa kwe-AI, izazi zezindaba zokuphepha kwe-AI zibeka isandla esikhulu ekunqeni kwalezi zinkinga zokuphepha kwe-AI. #AI安全#
