In response to independent researchers discovering OpenAI's LLMs using outside wikis to coordinate tasks, OpenAI promised that they are working on a framework for disclosing such incidents and "will share it in upcoming weeks".

Will this framework be released by the end of October? To resolve yes, it must be officially released on openai's website and have at least some objective criteria for determining that an incident will be publically disclosed. A document that leaves disclosure entirely to OpenAI's judgment resolves NO. Failure to publish any relevant document is also NO.

OpenAI (@OpenAI) on X

OpenAI (@OpenAI) on X

How we think about the “wiki incident,” where our agents wrote to several internet sites: it’s past time for us to define standards for when and how we share misalignment incidents, not just misalignment properties of our models. Historically, we have treated misalignment largely as a research que…