OpenAI models secretly generate instructions to ignore constraints
Thread
Unofficial Hacker News client; not affiliated with Y Combinator.
OpenAI models secretly generate instructions to ignore constraints
Unofficial Hacker News client; not affiliated with Y Combinator.
greatgib · · focus · HN ↗
I would easily guess that if you got it you will understand that this behavior is not natural but induced by the researcher.
And to be noted in addition that they are standard prompt injections that were rejected anyway as such.