Time Bandit ChatGPT jailbreak bypasses safeguards on sensitive topics
Time Bandit ChatGPT jailbreak bypasses safeguards on sensitive topics
I don't understand why the researcher needed to content the FBI to report this, just drop it in BugCrowd and call it a day. It's a ChatGPT jailbreak, not a Debian Zero-day.
I don't understand why we keep caring about this. AI "guard rails" will never be 100%. But beyond that everything is a google search away. Why does this matter?
everything is a google search away
Have you used Google search recently?
Fair enough. Everything is a searxng search away
So it'll tell you how to make weapons and stuff like that? The instructions to make a pipe bomb and a shotgun are freely available, and even if they weren't, it's not hard to figure out.
I don't understand his anxiety.
It's for the people who can't figure it out and the only thing preventing them from using the knowledge poorly are clear instructions.