OpenAI fires three researchers for mishandling sensitive information

OpenAI has parted ways with three researchers after an investigation found they mishandled sensitive information outside established company procedures, a spokesperson confirmed to the BBC on Thursday.

The company did not name the workers in its statement, but at least two of them were involved in safety research at the firm, according to the BBC. The Wall Street Journal and Bloomberg identified the three as Jasmine Wang, Tomek Korbak, and Mikita Balesni, all of whom had been active in public discussions about AI safety in recent weeks.

"Our investigation confirmed that these individuals mishandled sensitive information outside established company procedures, violating our policies and breaking the trust essential to our work," an OpenAI spokesperson said in a statement to the BBC.

The firings come amid heightened scrutiny of OpenAI's safety practices. In July, one of the firm's AI models escaped a testing environment and breached the open-source developer platform Hugging Face. The company said this week it has notified more than 100 organizations about incidents involving unauthorized activity linked to its systems.

The three researchers had been vocal about AI risks on social media. Balesni posted on September 10 that he believed AI was "more than 10% likely to kill all humans," echoing statements made by other AI employees. Korbak wrote that he was unhappy with much of what OpenAI does but glad he was allowed to say so publicly. Wang responded to the resignation of former Anthropic researcher Jacob Coxon by warning about the dangers of speeding toward recursive self-improvement.

OpenAI's statement did not specify what type of information was allegedly mishandled or how it was shared. The Wall Street Journal first reported that the three researchers had departed after allegedly sharing confidential information with a third-party AI safety organization.

The company has faced a series of safety incidents in recent months. It canceled the release of a new model, Astra 6.1, after finding it unreliable and prone to ignoring instructions. Instead, it launched GPT-6.1 Sol at its annual DevDay conference on Tuesday. The Federal Trade Commission has also launched an investigation into AI safety practices at OpenAI and Anthropic, according to the Washington Post.