Re:Prompt engineers have already jailbroken GPT-4 in just tw...
by rkangel on Friday March 17, 2023 @07:07AM
(#35195458)
That would work to a point. There is still a hole based on your trust of the underlying implementation. If you haven't read "Reflections on trusting trust" I recommend it (https://www.cs.cmu.edu/~rdriley/487/papers/Thompson_1984_Ref...).
[ Parent
| on HN ]