4 ms·
As proof, it gave me multiple rewrite answers saying that what i asked(random user’s ask) was not meeting their guidelines/safety and cannot answer.
by sturza 3y ago
As proof, it gave me multiple rewrite answers saying that what i asked(random user’s ask) was not meeting their guidelines/safety and cannot answer.
- wildrhythms 3y agoI don't understand how this is proof. OpenAI can easily have guardrails that prevent queries asking for other users information.
- sturza 3y agoWell, it's the best i got. I would expect it to hallucinate, but i would not expect it to hallucinate restricted content and then tell itself it's forbidden.