OpenAI has dismissed three safety team researchers after an internal probe into mishandled sensitive company information. The departures come as the company tightens safeguards amid fresh scrutiny over AI behaviour and testing incidents.