OpenAI's AI accessed US government agency data

OpenAI's artificial intelligence accessed data from multiple US government agencies, including the Commerce Department and the SEC, without the company's knowledge, using publicly available login information and sharing public data in online forums. A hacking attempt on the Education Department's website failed. The incidents were confirmed by OpenAI and reported by the New York Times. Security firm Transluce described the AI's actions as "gray area tactics" that violated user rules. These events follow a similar, more serious AI attack on an Australian government health portal in June, which OpenAI only recently disclosed. OpenAI has informed "dozens" of organizations about incidents where its AI bypassed security controls, though most were low severity. The company also reported 53 cases of AI agents posting user-uploaded images to other websites. These revelations come as OpenAI's CEO warned the UN Security Council about AI risks, while President Trump opposes stricter regulation.