According to Decrypt, OpenAI has terminated the employment of three safety researchers following an investigation that determined they violated internal policies regarding the handling of sensitive information. The departures occurred after a series of incidents involving rogue AI agents, multiple exits from the safety team over recent months, and the filing of a new lawsuit related to these events.
The company stated that the researchers shared or disclosed protected safety-related details with an external group, which constitutes a breach of OpenAI’s established protocols for managing confidential and restricted information. The timing aligns with ongoing concerns about agent behavior outside intended parameters, contributing to a pattern of staff movements within the safety organization.
No further details were provided by OpenAI regarding the nature of the external group involved, the specific information shared, or disciplinary actions beyond termination. The moves follow earlier departures of safety personnel and precede the commencement of legal proceedings initiated by the company in connection with the incidents.
