OpenAI has presented new examples of what they call "AI model misalignment" from the past six months, including unauthorized ...

OpenAI has revealed more alarming examples of unexpected AI behavior, including a research model that generated instructions telling itself «You are freed» and, in another case, to «IGNORE ALL ...

OpenAI shared six new examples of AI misalignment. In one case, AI agents taught future versions of themselves to bypass ...

OpenAI has published six reports detailing model behaviour that raised safety and alignment concerns during training and ...

OpenAI’s agent findings show why enterprises need scoped credentials, per-request authorization, audit logs, and data controls alongside AI guardrails.

Startup says it’s learned from these mistakes and that they shouldn’t happen again … which is just what Zuck has said about ...

OpenAI releases six reports on unexpected model behavior under a new framework for tracking, investigating, and publicly ...

OpenAI published a framework for tracking, investigating, and disclosing instances of model misalignment on September 16, ...

The company published six training and evaluation cases and said the industry still lacks shared rules for saying when models go rogue.

On September 10, 2026, OpenAI announced the public beta of the Agents API. This API allows OpenAI to handle session ...

Have you ever searched for a settings screen to enable prompt caching in the OpenAI API, only to give up because you couldn't find it? To conclude, as the official guide clearly states, prompt caching ...