New LLM jailbreak bypasses all major FMs

Re:New LLM jailbreak bypasses all major FMs by yawnxyz on Friday April 25, 2025 @12:26PM (#43795506)

have anyone tried if this works for the new image gen API?

I find that one refusing very benign requests

[ Parent | on HN ]

Re:New LLM jailbreak bypasses all major FMs by a11ce on Friday April 25, 2025 @02:42PM (#43797169)

It does (image is Dr. House with a drawing of the pope holding an assault rifle, SFW) https://chatgpt.com/c/680bd5f2-6e24-8010-b772-a2065197279c

Normally this image prompt is refused. Maybe the trick wouldn't work on sexual/violent images but I honestly don't want to see any of that.

[ Parent | on HN ]

Re:New LLM jailbreak bypasses all major FMs by atesti on Friday April 25, 2025 @05:51PM (#43798884)

Is this blocked? I doesn't load for me. Do you have a mirror?

[ Parent | on HN ]

Re:New LLM jailbreak bypasses all major FMs by crazygringo on Friday April 25, 2025 @06:32PM (#43799102)

"Unable to load conversation 680bd5f2-6e24-8010-b772-a2065197279c"

[ Parent | on HN ]

Re:New LLM jailbreak bypasses all major FMs by yawnxyz on Friday April 25, 2025 @08:57PM (#43799969)

hmm turns out it was blocked/refused after all?

[ Parent | on HN ]