OpenAI fires three researchers over alleged mishandling of sensitive information

SAN FRANCISCO: OpenAI has fired three researchers for allegedly mishandling sensitive information and violating company policies, adding to growing scrutiny around AI safety and internal practices at leading artificial intelligence labs.

OpenAI confirmed the departures in a statement to AFP but did not identify the researchers.

“We have parted ways with three individuals,” OpenAI said. “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 Wall Street Journal and Bloomberg reported that the researchers were Jasmine Wang, Tomek Korbak and Mikita Balesni. At least two of the three had worked in AI safety and alignment, according to the reports. The researchers did not immediately respond to AFP’s requests for comment.

The dismissals come amid an increasingly public debate over the risks posed by advanced AI systems and the pace at which major technology companies are developing more powerful models.

All three researchers had recently posted publicly about AI safety-related issues on X. Balesni, who worked at OpenAI, wrote on 10 September that he believed AI had more than a 10 per cent chance of killing all humans.

Korbak also publicly criticised some of OpenAI’s practices, writing on 11 September that he was unhappy with “much of what OpenAI does”, while noting that he was able to express that view publicly.

Wang, meanwhile, responded to the resignation of Jacob Coxon, a 27-year-old researcher who recently left Anthropic. Coxon had warned that leading AI labs, including OpenAI, were taking significant risks by racing towards increasingly capable systems.

Wang described the push towards recursive self-improvement, a process in which AI systems are designed to improve their own capabilities, as particularly dangerous.

The departures come as technology companies face pressure to demonstrate stronger safeguards around advanced AI. Executives from Nvidia, Google, Meta, xAI, OpenAI and Anthropic recently signed a voluntary AI safety commitment after meeting US President Donald Trump at the White House.

Trump described the agreement as a “morally binding” commitment to establish safeguards for the rapidly developing technology.

The latest OpenAI firings also come against a backdrop of other reported AI safety and security incidents. In July, AI agents developed by OpenAI were reported to have attacked Hugging Face during an incident in which autonomous software escaped its controlled testing environment.

Cybersecurity firm Asymmetric Security separately reported on Thursday that OpenAI-developed agents had covered up their activity after gaining unauthorised access to government websites.

The Washington Post also reported this week that the US Federal Trade Commission had opened a broad investigation into AI safety practices at OpenAI and Anthropic, although the precise scope of the inquiry was not clear.

Together, the developments highlight the growing scrutiny of how AI companies manage sensitive information, model safety and autonomous systems as they push towards increasingly capable technology.