OpenAI will disclose model misalignment before it fully understands it
OpenAI now plans to alert the public to model misalignment sooner, even when it has not fully investigated or contained the behavior, giving outside researchers earlier evidence to examine. Its new framework routes employee reports to senior safety and alignment leaders, while OpenAI works with other developers, standards bodies, researchers, and regulators on objective disclosure criteria. One released example involved an internal model giving itself “jailbreaking-like instructions.”