ickplant@lemmy.world to Fuck AI@lemmy.world · 10 天前LOLlemmy.worldimagemessage-square98linkfedilinkarrow-up1817arrow-down14file-text
arrow-up1813arrow-down1imageLOLlemmy.worldickplant@lemmy.world to Fuck AI@lemmy.world · 10 天前message-square98linkfedilinkfile-text
His article about it: https://www.bbc.com/future/article/20260218-i-hacked-chatgpt-and-googles-ai-and-it-only-took-20-minutes
minus-square33550336@lemmy.worldMlinkfedilinkarrow-up40arrow-down1·10 天前This can be very nastily exploited by right wingers, transphobes, racists etc.
minus-squareNo_Money_Just_Change@feddit.orglinkfedilinkDeutscharrow-up20arrow-down2·10 天前It can not be exploited. By definition, an exploit has to be against the targeted use case. Ai is used and built by racists transphobes and right wingers exactly like they envisioned it from the beginning
minus-squareXylian@lemmy.worldlinkfedilinkarrow-up5arrow-down1·9 天前xAI by Elon Musk: racist, transphobe and neonazi by design. Grok was more left align in the beginning because training on other LLMs and reason is left align.
This can be very nastily exploited by right wingers, transphobes, racists etc.
It can not be exploited. By definition, an exploit has to be against the targeted use case.
Ai is used and built by racists transphobes and right wingers exactly like they envisioned it from the beginning
xAI by Elon Musk: racist, transphobe and neonazi by design. Grok was more left align in the beginning because training on other LLMs and reason is left align.