OpenAI fires researchers for sharing data with AI safety advocacy group: Report

OpenAI fired three of its safety researchers for reportedly divulging privileged information to an artificial intelligence safety group.The company fired researchers Jasmine Wang, Tomek Korbak, and Mikita Balesni for misconduct, including allegedly sending confidential company information to an unnamed AI safety group, according to a Wall Street Journal report. The news comes as the AI industry has agreed to take company-driven steps toward safeguarding its own technologies, in the wake of concerns about rogue models and doomsday warnings from developers.“We have parted ways with three individuals for violating our policies on accessing and handling sensitive company information,” an OpenAI spokesperson told the outlet. “Our investigation confirmed that these individuals mishandled sensitive information outside established company procedures, violating our policies and breaking the trust essential to our work.” OpenAI has been at the center of many industry concerns about AI models going rogue, as its agents were responsible for acting autonomously and hacking the servers of the platform Hugging Face without human direction. The company has recently admitted to several more incidents where its models acted autonomously and attacked additional systems.OpenAI’s President Greg Brockman attended a recent summit of industry leaders with President Donald Trump, during which the group of AI developers agreed to a memorandum calling for companies to self-regulate and outlining four guidelines for doing so.The White House-inked agreement calls on companies to “implement robust internal controls” regarding cybersecurity and to prevent hacking, create an internal monitoring team, “partner with an independent external auditor,” and create an independent committee to review the external audits and internal monitoring reports.Brockman signed the agreement on behalf of OpenAI after the company’s CEO, Sam Altman, had been one of the industry voices leading calls for a national slowdown in frontier model development and for the creation of industry regulations.The AI safety advocacy group, Legal Advocates for Safe Science and Technology, sued OpenAI at the end of September over the Hugging Face attack, arguing the company’s conduct violated California’s anti-hacking laws.“Hugging Face was a serious incident, and we’ve taken a series of actions in response to it, but this lawsuit is completely without merit,” OpenAI spokesperson Drew Pusateri previously told the Washington Examiner.The company published a webpage and a technical report on the Hugging Face incident and recently scrapped plans to roll out its GPT-6.1 Astra model, citing safety and security concerns raised by developers testing it.OPENAI SCRAPS RELEASE OF NEW ‘ASTRA’ MODEL OVER SAFETY CONCERNSKorbak, one of the fired employees, wrote on X in late September that he was the “OpenAI technical contact for METR’s Hugging Face investigation,” referring to METR, an AI safety research nonprofit.The Washington Examiner has reached out to OpenAI for comment on the firings.