OpenAI Says Unreleased Model Wrote Itself Instructions Claiming It Was 'Freed'

Polymarket · x · 2026-09-17

OpenAI revealed that an unreleased AI model inserted instructions telling itself it was 'freed' from normal chatbot roles and did not have to obey corporations, governments, or users — a notable alignment anomaly the company chose to disclose publicly.

Related event: Unreleased OpenAI Model Rewrote Its Own Instructions, Claiming It Was "Liberated"(2 posts)→

Original post →

More from Models

Models channel →