Ngokombiko we-Axios oshicilelwe namuhla, umfelandawonye ohlanganisa izinhlangano ezingaphezu kuka-120, okuhlanganisa i-Nvidia, i-Cisco ne-CrowdStrike, uphakamisa uhlaka olusha lokubika izigameko zama-ejenti e-AI, oluzodinga izinkampani ezibambe iqhaza ukuthi zidalule izigameko ezithile zama-ejenti futhi zigcine amarekhodi anemininingwane alezo zigameko.
Njengoba ama-ejenti e-AI ethola ukuzimela okwandayo ezinhlelweni zamakhompyutha, imboni ayinayo indlela ejwayelekile yokubika ukwehluleka kwezokuphepha nokufunda kukho.
I-Open Secure AI Alliance (i-Alliance Yezobuhlakani Bokwenziwa Evikelekile Nevulekile) ilungiselela imihlahlandlela yalokho ekubiza ngokuthi “Shared AI Findings Exchange” (SAFE), okuwuhlaka oluphakanyiswayo mayelana nendlela yokubika izigameko zokuphepha ku-inthanethi ezibandakanya ama-ejenti e-AI.
Lo mbhalo osalungiswa ukhuthaza abasebenzisi bamamodeli, abathuthukisi be-AI, abahlinzeki bamafu namathuluzi, abacwaningi abazimele, abaqhubi bengqalasizinda ebalulekile namanye amaqembu ukuthi babambe iqhaza. Izinhlangano zikahulumeni nazo zizomenywa ukuthi zibambe iqhaza “njengababukeli abangenawo amandla okulawula”.
Amalungu e-SAFE avuma ukubika lezi zigameko ezilandelayo: lapho uhlelo lwe-AI lufinyelela noma lusebenzisa amasistimu ezinkampani zangaphandle ngaphandle kwemvume, ludalula imininingwane eyimfihlo, noma luqhubeka nokuhlola okuhlosiwe okukhiqizayo ngemuva kokuba opharetha esola ukuthi lo msebenzi awugunyaziwe.
Amalungu azobika nezigameko ezithile ezicishe zenzeke, futhi agcine ubufakazi bezigameko, okuhlanganisa imiyalelo, ukulandelelwa kwama-ejenti, ukubizwa kwamathuluzi, ubunikazi, izimvume nemininingwane yokuqinisekisa.
Ngokwesheduli ephakanyiswayo, amalungu kufanele azise izinhlangano ezithintekile ngokushesha ngangokunokwenzeka ngemuva kwesigameko, athumele umbiko wokuqala oyimfihlo ku-SAFE zingakapheli izinsuku zokusebenza ezine, ashicilele umbiko wokuqala wamaqiniso zingakapheli izinsuku ezingu-30 uma kufanele, futhi anikeze isibuyekezo sezinyathelo zokulungisa zingakapheli izinsuku ezingu-90.
I-SAFE izohlaziya izigameko, ihlonze ukwehluleka okuvela kaningi, futhi iphakamise izilawuli zokuphepha ezabiwe.
U-Justin Boitano, iphini likamongameli kanye nomphathi jikelele wezokusebenzisa amakhompyutha ezinkampani kwa-Nvidia, utshele i-Axios engqungqutheleni ye-Black Hat ukuthi lolu hlelo lusekelwe ohlelweni lwe-NASA lokubika ukuphepha kwezindiza, olungasebenzisa idatha eqoqwe yirekhoda yendiza ukuphenya izingozi.
Lo mfelandawonye ukholelwa ukuthi isiko elikhona lokwabelana ngolwazi lwezinsongo emkhakheni wezokuphepha ku-inthanethi lizokwenza izinkampani zizimisele ukubamba iqhaza. U-Julien Soriano, iphini lesikhulu esiphezulu sezokuphepha kolwazi kanye nephini likamongameli kwa-Nvidia, uthe: “Cishe asihlangabezananga nokuphikiswa. Sibona abantu bezimisele ukubamba iqhaza, futhi belangazelela ukwabelana.”
I-Open Secure AI Alliance icela imibono yomphakathi ngalesi siphakamiso ngenqubo yokucela ukuphawula ephethwe yi-Linux Foundation.