@[email protected]MB to Hacker [email protected]English • 9 months agoAI poisoning could turn open models into destructive "sleeper agents"arstechnica.comexternal-linkmessage-square2fedilinkarrow-up110arrow-down12file-text
arrow-up18arrow-down1external-linkAI poisoning could turn open models into destructive "sleeper agents"arstechnica.com@[email protected]MB to Hacker [email protected]English • 9 months agomessage-square2fedilinkfile-text
minus-square@DadiferlinkEnglish2•edit-29 months agoThat’s pretty terrifying. I’m not sure why open source language models would be more vulnerable to this than closed source, However.
That’s pretty terrifying. I’m not sure why open source language models would be more vulnerable to this than closed source, However.