I-European Union yethule uchungechunge lwemithetho emisha okuhloswe ngayo ukusiza umphakathi ukuba uhlonze kalula ama-chatbot nokuqukethwe kwe-AI okuyi-deepfake ku-inthanethi. Izimfuneko ezintsha zokubonisa obala, eziyingxenye ebalulekile yoMthetho we-European Union Wobuhlakani Bokwenziwa (AI Act), ziqale ukusebenza ngokusemthethweni ngomhlaka-2 Agasti. Zidinga ukuthi izinkampani zidalule ukuthi abasebenzisi bayasebenzisana yini nemodeli ye-AI, nokuthi okuqukethwe okuthintekayo kukhiqizwe noma kwaguqulwa yi-AI yini.
Le mithetho yokubonisa obala ibeka izimfuneko ezihlukene kubabambiqhaza abahlukene, okuhlanganisa abahlinzeki bezinhlelo ze-AI (izinkampani ezakha futhi ezithengisa izinhlelo ze-AI) kanye nabasebenzisi bezinhlelo (amapulatifomu nezinsizakalo ezisebenzisa lezi zinhlelo ze-AI). Nokho, ezinye izinkampani zingena kuzo zombili lezi zigaba, njenge-Meta ne-SpaceXAI.
Kubaphakeli, kufanele bakhe izinhlelo ze-AI ngendlela ezokwazisa ngokusobala abasebenzisi lapho besebenzisana ne-AI kunomuntu wangempela, “ngaphandle kwalapho lokhu kusobala”. Ngaphezu kwalokho, izinkampani kufanele zifake izimpawu ezifundeka ngomshini kumsindo, ezithombeni, kumavidiyo nasembhalweni okukhiqizwe yi-AI, ukuze kutholakale ukuthi kukhiqiziwe noma kwaguqulwa ngokuzenzisa.
Ngesikhathi esifanayo, abasebenzisi bezinhlelo kufanele babhale okuqukethwe okuyi-deepfake kwe-AI okukhiqiziwe noma okuguquliwe, okuhlanganisa izithombe, umsindo namavidiyo okuhloswe ngawo ukubukeka njengokwangempela.
IKhomishini yaseYurophu ithe emhlahlandleleni wayo wokubonisa obala: “Ukuthuthuka okusheshayo kwezinhlelo ze-AI ezikhiqizayo nezisebenzisanayo kwenza kube nzima kakhulu kubantu ukuhlukanisa phakathi kokusebenzisana ne-AI nokuqukethwe okukhiqizwe yi-AI, nokuqukethwe kwangempela okudalwe abantu.”
“Abantu kufanele bazi ukuthi basebenzisana nini ne-AI, noma ukuthi babhekene nini nokuqukethwe okukhiqizwe yi-AI. Lokhu kuzobasiza ukuba benze izinqumo ezinolwazi olungcono, balungise izinga lokwethemba kwabo i-AI nendlela abathembele ngayo kuyo, futhi bagweme ukuthintwa ulwazi olungelona iqiniso noma ukukhohliswa.”
IKhomishini yaseYurophu iphinde yenza iqoqo lamalebula okudalula okuqukethwe kwe-AI, amapulatifomu ezobuchwepheshe angawasebenzisa ngokwezidingo zoMthetho Wobuhlakani Bokwenziwa, futhi yanikeza nezibonelo zezimo ezithile zokusetshenziswa.
Lawa malebula ahambisana nezimpawu ezifanayo ezethulwe ngaphambilini amapulatifomu afana ne-TikTok, Instagram ne-Facebook. I-European Union yaklama lezi zithonjana ezifanayo ukuze igweme ukuthi amapulatifomu ahlukene akhe amalebula awo ahlukene. Nakuba izithonjana eziyizibonelo ezinikezwe yi-European Union kungaphoqelekile ukuzisebenzisa, iKhomishini yaseYurophu ikubeke kwacaca ukuthi izimfuneko zayo zokubhala okuqukethwe kwe-AI “akuzona izinto ongakhetha ukuzenza noma ungazenzi”.
Izinkampani ezihluleka ukuthobela le mithetho emisha yokubonisa obala zingabhekana nenhlawulo efinyelela kuma-euro ayizigidi ezingu-15 (inothi: ngezinga lokushintshisana lamanje, cishe ama-yuan ayizigidi ezingu-117) (cishe ama-dollar aseMelika ayizigidi ezingu-17.2 (ngezinga lokushintshisana lamanje, cishe ama-yuan ayizigidi ezingu-116)), noma inhlawulo elingana no-3% wemali engenayo yazo yonyaka emhlabeni wonke.
Kuzinhlelo ezintsha ze-AI ezethulwe ngemva komhlaka-2 Agasti, le mithetho yokubonisa obala izosebenza ngokushesha. Amamodeli nezinsizakalo ezazivele zethulwe ngaphambi komhlaka-2 Agasti zinikezwe isikhathi soshintsho esiyizinyanga ezi-4, futhi kufanele zenze izinguquko zokuhambisana nomthetho ngaphambi komhlaka-2 Disemba.