Three fired OpenAI safety researchers say the company got the story wrong, and that the firings are already scaring their former colleagues into silence.
OpenAI dismissed the three researchers this week, citing mishandling of sensitive information. In an open letter, the researchers dispute that account, rejecting the specific misconduct allegations against them. They argue the dismissals misrepresent what actually happened and warn the episode is reshaping how people inside OpenAI's safety function now behave. The letter frames the firings as a signal to anyone still weighing whether to raise a safety concern.
That is the real stakes: not whether three people broke a rule, but whether OpenAI's internal safety process can work if flagging a problem carries firing risk. A safety team is only useful if researchers trust they can escalate issues without jeopardizing their jobs. A disputed firing, fought out in public rather than settled quietly, suggests that trust is already cracking.
Nobody outside OpenAI has seen the underlying evidence, so the company's "misconduct" label and the researchers' rebuttal both deserve equal skepticism for now.