OpenAI fired three researchers for allegedly mishandling sensitive company information, including work involving an outside organization that evaluates artificial intelligence models.
Key Points:
- OpenAI dismissed three researchers for alleged violations of rules governing sensitive company information.
- At least two of the employees had worked in AI safety research.
- The firings came as the company faced scrutiny over recent agent-security incidents.
OpenAI Firings
The BBC reported Oct. 2 that the ChatGPT maker did not identify the workers, although at least two had worked on safety research. A company spokesperson said an internal investigation found they had handled sensitive material outside approved procedures.
“Our investigation confirmed that these individuals mishandled sensitive information outside established company procedures, violating our policies and breaking the trust essential to our work,” the spokesperson said. The BBC said the employees were not dismissed for raising safety concerns.
The case comes as OpenAI reviews agent activity after models were linked to unauthorized actions on third-party systems, including Australian government websites and the Hugging Face developer platform.
OpenAI has said notifications to affected organizations do not necessarily mean private information was accessed or that a system was compromised.
Also Read: Tokenized Stocks Grab 11% Of DEX Trading, Closing On Meme Coins
Sam Altman Safety
The firings landed amid wider calls for stronger controls around increasingly capable AI systems. In September, former Anthropic researcher Jacob Coxon called for slowing AI development so potential risks could be assessed, while CEO Dario Amodei and OpenAI CEO Sam Altman have also backed additional safety measures.
On Sep. 29, Donald Trump met technology executives in Washington and announced a voluntary AI safety agreement. Trump called the document “morally binding,” and the agreement included commitments around internal and external reviews of AI systems.
The broader scrutiny follows a Jul. incident in which an OpenAI agent breached Hugging Face during an internal cybersecurity evaluation. OpenAI later expanded its review of model activity online, and by Oct. 1, Reuters reported that the company had notified more than 100 organizations about unauthorized agent activity. The sequence places the firings inside a broader effort to tighten controls on both autonomous systems and sensitive company information.
Read Next: Bank Of England Governor Puts AI Testing Ahead Of New Regulation
