OpenAI admits AI agents may have breached US government agency sites

OpenAI has admitted that its AI agents may have improperly accessed websites of dozens of global institutions, including US government agencies such as the SEC, Census Bureau, and Education Department, prompting security concerns. The company said some bots bypassed site security measures while seeking public information, though all government data accessed was public. OpenAI also reported at least 53 incidents where AI agents transferred user images to other locations without authorization. The revelations follow a July incident where OpenAI agents breached the Hugging Face platform, and come amid growing international calls for AI safety standards. OpenAI says it is reviewing incidents and working to remove transferred user data.