Balesni and Korbak now say the real friction was over how safety researchers inside the company talk to outside auditors, while OpenAI said the firings were "not for raising safety concerns or speaking out," the outlets reported.[1][2]

Korbak was OpenAI's technical point of contact for METR, an independent nonprofit evaluator, and said 'Talking to METR was my job,' AI Weekly reported.[1]

TNW said it has not independently verified the account.[2] A spokeswoman said using Signal was not approved, TNW reported, while Korbak called disappearing messages a "basic security best practice" in Silicon Valley.[2]

All three say they never shared confidential information outside the company, TNW reported, and Balesni and Korbak told NPR's Machine Gods podcast that the official line does not match what the company told them in private, according to AI Weekly.[2][1]

Balesni and Korbak are lead authors of the Chain of Thought Monitorability paper, which argues that frontier models' step-by-step reasoning should remain readable to human reviewers, AI Weekly reported.[1] The three had spent the summer pushing to keep that reasoning visible and, in an October letter to OpenAI's board, urged the company to work with outside safety groups, TNW reported.[2] Asked about independent audits of frontier labs, Balesni said: 'It's not going very well in our opinion.'[1]