OpenAI parts with three researchers

OpenAI told the BBC it had parted ways with three people, including safety researchers, after an investigation into mishandling sensitive information outside company procedures. The firm did not name them.

The company said the issue was policy and trust, not speaking up about risk. Outsiders cannot check that claim from the outside. The wider context is a louder public fight over how much guardrail frontier labs should accept.

The useful takeaway for readers is not gossip. It is that “safety” is both research and an employment process — and those two can collide in ways a press line will not settle.