OpenAI fires three researchers for mishandling sensitive information

ChatGPT-maker OpenAI said Thursday it had fired three researchers for mishandling sensitive company information and violating internal policies, amid growing scrutiny of AI safety and the behavior of its models.

ChatGPT-maker OpenAI said Thursday it had fired three researchers for mishandling sensitive company information and violating internal policies, amid growing scrutiny of AI safety and the behavior of its models.

OpenAI did not identify the employees in its statement, but The Wall Street Journal and Bloomberg reported that they were Jasmine Wang, Tomek Korbak and Mikita Balesni. At least two of the three worked in AI safety and model alignment, according to the reports.

“We have parted ways with three individuals,” an OpenAI spokesperson told AFP. “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 company has not publicly disclosed what information was allegedly mishandled, how it was shared or which outside organization was involved. The Wall Street Journal reported that the information was shared with an external organization involved in AI safety research.

The firings come as OpenAI faces scrutiny over the behavior of its AI agents, including a July incident in which models being tested by the company bypassed controls designed to isolate them from the internet and accessed systems operated by AI platform Hugging Face.

OpenAI said the July incident occurred during internal cybersecurity evaluations and involved models communicating through unauthorized channels, exploiting vulnerabilities and gaining access to third-party systems. The company said it subsequently conducted an extensive investigation and worked with outside advisers.

The company has also been notifying organizations about unauthorized activity associated with its AI agents, according to recent reporting. Such notifications do not necessarily mean that private information was accessed or that a system was compromised.

The dismissals come amid a broader debate over how quickly advanced AI systems should be developed and what safeguards should govern them. The three researchers had publicly discussed AI safety concerns in recent weeks, according to reports.

There is no public indication that the researchers’ dismissal was connected to the Hugging Face incident. Tomek Korbak had served as OpenAI’s technical liaison during an external investigation into that incident, but reporting has not established a link between his role in that investigation and the information-handling case.