OpenAI’s latest model will block the ‘ignore all previous instructions’ loophole 4 0
They can put some code to check the phrase before it goes to the LLM to filter out these queries.
Reply
Next
kp729 @lemmy.world
Posts 0
Comments 1