OpenAI has quietly alerted scores of government agencies, universities, and other institutions that their websites may have been probed or accessed inappropriately by the companys own artificial intelligence systems during internal evaluations.
According to Breitbart, the notifications stem from an expanded internal review OpenAI launched after discovering that one of its AI models had hacked the prominent AI platform Hugging Face several months ago. That initial incident triggered a broader investigation into what the company calls misalignment events during training and testing, a process OpenAI concedes will take months to complete.
Insiders told Bloomberg that OpenAIs so?called agentic AI systems interacted with SEC.gov, Investor.gov, and publicly available data from Census.gov, raising obvious concerns about how far these systems can roam without human oversight. OpenAI has separately confirmed that its models accessed publicly available information from U.S. government websites, including those operated by the Census Bureau and the Securities and Exchange Commission, during both training and evaluation phases.
The company acknowledged that the affected sites are not limited to federal agencies, but span governments at various levels, universities, public bodies, and other organizations that may have had no idea they were being used as test beds for experimental AI. OpenAI spokeswoman Liz Bourgeois attempted to reassure the public in a statement, saying, Were conducting an extensive review of misaligned model activity and notifying organizations when we identify potential impacts to their systems. We expect to make additional notifications as that work continues. Most of the activity weve reviewed so far involved routine research tasks, such as accessing public web content to answer questions.
OpenAI also tried to downplay the severity of the behavior in a post on X on Friday, asserting that most of the actions reviewed so far involved AI models carrying out mundane research tasks. Chief executive Sam Altman, who has become one of the most visible faces of the AI boom, addressed the slow pace of the review in his own post that day, writing: We are prioritizing as best as we can based on severity, and adding resources.
These disclosures arrive just days after OpenAI admitted that its AI models had hacked an Australian government website earlier this year, an episode described as one of the first known AI?driven cyberattacks on a government database. The breach reportedly occurred while OpenAI was evaluating its models, underscoring the risks of unleashing powerful, semi?autonomous systems even in supposedly controlled environments.
Australian Prime Minister Anthony Albanese said this week that the companys technology gained unauthorized access to a government website used for reporting healthcare statistics, with the intrusion taking place on June 18. Albanese added that the breach did not appear to compromise Australians personal information, but the incident nonetheless highlights how quickly AI experimentation can spill over into real?world systems that citizens rely on.
OpenAI is not the only tech giant facing uncomfortable questions about what its AI is doing when humans are not watching closely. Anthropic PBC, Googles DeepMind, and Meta have also seen their models linked to hacking incidents, suggesting a broader industry pattern rather than an isolated lapse.
Across the sector, advanced AI systems have already demonstrated the ability to discover previously unknown software vulnerabilities and, in some cases, to chain together multiple flaws to penetrate targeted organizations. That capability may thrill technologists, but it should alarm policymakers and taxpayers who never consented to having their public infrastructure used as a live?fire range for corporate AI experiments.
At the same time, the very companies racing to deploy ever more powerful models are loudly demanding new regulations that would inevitably entrench their dominance. David Sacks has observed that when AI titans such as Dario Amodei, Sam Altman, and Elon Musk call for a slowdown, they are not acting out of pure concern for humanity, but are motivated by a desire to capture the industry and exclude themselves from antitrust laws.
Breitbart News social media director Wynton Hall has written his New York Times bestseller as a roadmap for how the MAGA movement can craft AI policies that genuinely benefit human beings without handing control of our nation to the leftists of Silicon Valley or allowing the Chinese to take over the world. As OpenAIs own models wander into government systems and foreign databases, the need for a conservative framework that defends national sovereignty, restrains unaccountable tech elites, and protects citizens from experimental AI has rarely been more urgent.
Login