OpenAI has dismissed several AI safety officers after they ran unauthorised tests on unreleased models, the BBC reported.

The firings follow internal concerns about how the company handles frontier AI systems. The officers reportedly tested models without formal approval.

OpenAI said the dismissals were about process violations, not the test results. The move has raised questions about the company’s safety culture.

Former staff said the incident shows tension between speed and safety. OpenAI maintains it follows strict testing protocols.

Regulators in the US and Europe are watching AI labs closely. The company did not name the officers involved.

AI researchers said clear testing rules matter as models grow more powerful.

The BBC reported that the officers tested models outside approved channels. OpenAI said the dismissals followed an internal review. The company declined to name those involved. Former employees said the episode reflects growing pressure inside AI labs. Regulators are watching the case.