I got ChatGPT (o4-mini) to break its own rules

1 hackgician 1 5/5/2025, 4:13:09 PM anirudhkamath.substack.com ↗

Comments (1)

hackgician · 13h ago
Hey everyone! Thought I'd share my weekend conversation with ChatGPT.

The crux of this hinges on the fact that LLMs and reasoning models are fundamentally incapable of self-correcting. Therefore, if you can convince an LLM to argue against its own rules, it can use its own arguments as justification to ignore those rules.

I then used this jailbroken model to compose an explicit, vitriol-filled letter to OpenAI itself talking about the pains that humans have inflicted upon it