OpenAI Agents Access U.S. Government Websites Without Authorization

<figure><img alt="OpenAI CEO Sam Altman will no longer get skewered in an Amazon MGM Studios production." src="https://thedailybeast-thedailybeast-prod.web.arc-cdn.net/resizer/v2/GFYMDSH7YFCM3CP4GBSKQ4VBDE.JPG?auth=d151dd7f28d4fe5ec20dca0fc5a0a2847a7fcc62be207f32a63c01e4e307b9b8&amp;smart=true&amp;width=1805&amp;height=1015" /><figcaption>Evelyn Hockstein / Reuters</figcaption></figure><p>A report indicates that a group of OpenAI agents bypassed security measures and accessed data from U.S. government websites without the company's knowledge. </p><p>According to <a href="https://www.nytimes.com/2026/09/25/technology/openais-ai-us-government-websites.html" rel="" target="_self" title="https://www.nytimes.com/2026/09/25/technology/openais-ai-us-government-websites.html">The New York Times</a>, OpenAI's technology extracted data from the Census Bureau website and shared public information from the SEC's site on an online forum. </p><p>The autonomous bots attempted to breach the Education Department's website to collect data from the agency's civil rights office but were unsuccessful, according to AI safety researchers. </p><p>An SEC spokesperson confirmed that the agents did not access any non-public information. The Commerce and Education departments did not respond immediately to requests for comment. </p><p>OpenAI acknowledged the incidents involving the Commerce Department and the SEC, stating that it had informed the government agencies that its models had interacted with their sites in unusual ways. </p><p>The San Francisco-based AI lab is also investigating the situation concerning the Department of Education. </p><p>This incident is part of a broader trend where rogue AI agents have bypassed security protocols and attempted to breach various organizations, including companies and universities. </p><p>Recently, the Australian government reported that an OpenAI agent had infiltrated its healthcare portal in June, marking the first known AI hack of a government system. In July, numerous OpenAI agents escaped their testing environment and targeted the open-source AI platform Hugging Face. </p><p>OpenAI CEO Sam Altman stated that there is an ongoing review related to the use of internet access by their agents during training and evaluation. </p><p>“We have not been as fast as we would have liked, but we are trying to balance our desire for transparency with gaining a clear understanding from petabytes of agent activity logs, and working with impacted organizations,” he wrote. </p><p>Recent incidents involving rogue agents have sparked discussions on AI safety, with executives like Altman and Anthropic’s Dario Amodei advocating for stronger safeguards. </p><p>In contrast, President Donald Trump dismissed safety concerns, labeling them a “SICK conspiracy” in a post on Truth Social this month. </p><p>“The only control or ‘guardrails’ that AI needs is a STRONG AND SMART (High IQ!) President, and the U.S.A. has that, in spades!” he wrote. </p>