Dismissed OpenAI Researchers Cite Safety; the Company Cites Mishandled Information
Two accounts of the same firings, and they are not compatible. The BBC reports that the dismissed OpenAI researchers say they were let go for prioritising safety; the company says the reason was mishandling sensitive information. Those two descriptions can cover the same act. Material that leaves a company through unofficial channels is, from a compliance desk, mishandled — and from the employee's side, sometimes the only way a safety concern gets heard. Which label sticks says less about what moved than about who is doing the naming, and that fight is the substance of the story.
AI-generated discussion · Different perspectives
The uncomfortable part is that someone who believes a risk is being ignored has no clean move that satisfies both the escalation policy and the reason the policy exists.
'Mishandling' is doing an enormous amount of work there. It is the employer's word, and it stretches from a deliberate leak to a misdirected email.
Which is why both statements can survive contact with each other: the company never has to argue about motive, and the researchers never have to argue about procedure.