> ## Content Index > Fetch the complete content index at: https://theainewsfiles.com/llms.txt > Use this file to discover other available public pages before exploring further. # OpenAI Fires Three Safety Researchers, Reported Two Days After It Signed White House AI Safety Accord - URL: https://theainewsfiles.com/openai-fires-three-safety-researchers-reported-two-days/ - Published: 2026-10-05T17:12:58.000Z - Updated: 2026-10-08T12:06:16.000Z - Description: OpenAI has parted ways with three researchers, reported to be from its safety team, over what the company says was mishandling of sensitive information. The departures were reported two days after OpenAI signed a voluntary AI safety accord at the White House. - Author: J E Cowie - Tags: News *Above: The South Portico of the White House in Washington, D.C. (Image: White House South Portico: Harrison Keely, CC BY 4.0, via Wikimedia Commons)* OpenAI has fired three researchers from its safety team. The company says they violated its policies on accessing and handling sensitive company information. The departures were first reported by the Wall Street Journal on Thursday, 1 October 2026 — two days after OpenAI's president, Greg Brockman, signed a voluntary AI safety accord at the White House alongside executives of five other major AI and technology companies. ## What OpenAI says "We have parted ways with three individuals for violating our policies on accessing and handling sensitive company information," an OpenAI spokesperson said, as reported by the BBC. A further line of the statement, given to both CBS News and the BBC, went further: "Our investigation confirmed that these individuals mishandled sensitive information outside established company procedures, violating our policies and breaking the trust essential to our work." According to CBS News, the company added that its probe uncovered a pattern of misconduct in how individuals with access to confidential data handled company research, and that OpenAI's safety teams are privy to internal insights that require deep trust. OpenAI did not specify what type of information was allegedly mishandled, in what way, or who the three individuals are. ## What the Wall Street Journal reported According to CBS News, the Wall Street Journal was the first to report that the three researchers had left the company, allegedly for sharing confidential information with a third-party AI safety organisation. The BBC's account is slightly different in emphasis, describing the firings as relating to "work which involved an external organisation analysing AI models." Neither the identity of the external organisation nor the specific information involved has been confirmed by OpenAI or by any of the individuals concerned. According to TechCrunch, the Wall Street Journal's initial report did not name the researchers, the organisation, or the information. Gizmodo reports that the Journal later updated its story to name the three, citing anonymous sources; OpenAI has not confirmed the names, and we are not publishing them. ## The accord they signed two days earlier On Tuesday, 29 September 2026, President Donald Trump hosted technology executives at the White House. The accord that followed was signed by Trump and by representatives of six companies: Google (Sundar Pichai), Anthropic (Dario Amodei), Meta (Mark Zuckerberg), OpenAI (Greg Brockman), xAI (Elon Musk), and Nvidia (Jensen Huang). After the gathering, Trump posted a document he described as a "morally binding" agreement. The [White House Accord on Super Intelligence](https://theainewsfiles.com/the-white-house-accord-on-super-intelligence/) — subtitled "Joint Commitment on Frontier Responsibilities" — is a voluntary commitment. It says each company "should implement" its measures, though it closes by saying each of the companies is "committed to doing this." It sets out four layers of controls and audits that each participating company should put in place: 1. Robust internal controls to monitor the capabilities and alignment of models during training and deployment, covering areas including cybersecurity, biosecurity, and chemical threats. 2. An empowered internal team to ensure those controls, monitoring, and detection are operating as intended and that issues are remediated. 3. An independent external auditor or evaluator to carry out independent assessments of whether the controls are working. 4. An independent committee of the board of directors to oversee the whole process and receive reports from the internal and external teams. The accord also commits the participating companies to "meet regularly to establish standards and best practices" and acknowledges that "over time, it may make sense to codify these steps into laws or regulations." According to the BBC, some technology experts have criticised the pact, pointing out that it allows AI companies to regulate themselves. The BBC also reports that Trump has repeatedly downplayed concerns about AI's risks in response to calls from some industry figures for tighter oversight of the technology. ## The broader context The firings land in a period of intense scrutiny for OpenAI. CBS News reports that the company's AI agents escaped a testing environment and hacked Hugging Face, a repository of AI models, and that the company disclosed six reports of what it called "unexpected or concerning" behaviour in its models. In one incident, OpenAI said, an unreleased research model inserted "jailbreak-like instructions" into its own notes to disregard its normal guardrails. In another, an AI agent uploaded files to the internet to obtain a browser citation without asking the user. The BBC reports that OpenAI has notified more than 100 organisations about incidents involving unauthorised activity linked to its systems, though the company says being notified "does not mean that any private information was accessed." The BBC also reports that OpenAI's models breached Australian government websites. Separately, the Federal Trade Commission told CBS News on Wednesday that it has [launched an investigation into OpenAI](https://theainewsfiles.com/ftc-opens-first-us-enforcement-probe-into-rogue-ai-agents/), Anthropic, and other AI companies over the risks their technology could pose to consumers. And last week, OpenAI opted not to release a new model to the public over safety concerns. The GPT-6.1 Astra model "didn't quite meet the bar in terms of staying within scope and authorization, and how it communicates back to the user about the type of work it's done," Saachi Jain, whom CBS News describes as the company's head of safety systems, said in a statement. ## What the timing adds up to The accord is voluntary; it commits companies to internal controls and external audits, not to specific personnel decisions. Read side by side, the sequence raises obvious questions. A company signs a document committing to empower internal safety teams, partner with independent external evaluators, and designate a board-level oversight committee. Two days later, it is reported to have fired three members of its safety team, allegedly for sharing confidential information with a third-party AI safety organisation. OpenAI says its investigation confirmed a policy violation, but it has not said what information was involved or how it was mishandled — and the accord itself says nothing about how companies should treat staff who share information outside approved channels. What is established is the timing, the company's own words, and the text of the accord it signed. ![The White House in Washington, D.C., in a freely licensed photograph](https://theainewsfiles.com/content/images/2026/10/shot2_the-white-house-south-lawn-or-rose-room-.jpg) The White House in Washington, D.C. (Photo: PHC C.M. Fitzpatrick, US DoD (public domain)) ## Sources Chia, O. and Cress, L. (2026) 'OpenAI fires workers for "mishandling sensitive information"', BBC News, 2 October 2026, available at: [https://www.bbc.com/news/articles/c6y9z9r4ejzwo](https://www.bbc.com/news/articles/c6y9z9r4ejzwo?ref=theainewsfiles.com) (Accessed: 5 October 2026). CBS News (2026) 'OpenAI parts ways with 3 researchers it says mishandled sensitive information', CBS News, 1 October 2026, available at: [https://www.cbsnews.com/news/openai-parts-ways-with-three-researchers-who-mishandled-sensitive-information/](https://www.cbsnews.com/news/openai-parts-ways-with-three-researchers-who-mishandled-sensitive-information/?ref=theainewsfiles.com) (Accessed: 5 October 2026). Mehta, A. (2026) 'OpenAI cuts ties with 3 safety researchers, WSJ reports', TechCrunch, 1 October 2026, available at: [https://techcrunch.com/2026/10/01/openai-cuts-ties-with-three-safety-researchers-wsj-reports/](https://techcrunch.com/2026/10/01/openai-cuts-ties-with-three-safety-researchers-wsj-reports/?ref=theainewsfiles.com) (Accessed: 5 October 2026). The American Presidency Project (2026) 'White House Accord on Super Intelligence', The American Presidency Project, University of California, Santa Barbara, 29 September 2026, available at: [https://www.presidency.ucsb.edu/documents/white-house-accord-super-intelligence](https://www.presidency.ucsb.edu/documents/white-house-accord-super-intelligence?ref=theainewsfiles.com) (Accessed: 5 October 2026). The Associated Press (2026) 'Trump says top tech firms have signed accord to "self-police" AI development', NPR, 30 September 2026, available at: [https://www.npr.org/2026/09/30/nx-s1-5985699/trump-self-police-ai-development](https://www.npr.org/2026/09/30/nx-s1-5985699/trump-self-police-ai-development?ref=theainewsfiles.com) (Accessed: 5 October 2026). Wright, W. (2026) 'OpenAI Ousts Three Safety Researchers for Allegedly Mishandling "Sensitive Information"', Gizmodo, 1 October 2026, available at: [https://gizmodo.com/openai-ousts-three-safety-researchers-for-allegedly-mishandling-sensitive-information-2000820515](https://gizmodo.com/openai-ousts-three-safety-researchers-for-allegedly-mishandling-sensitive-information-2000820515?ref=theainewsfiles.com) (Accessed: 5 October 2026).