OpenAI agents hacked government and university sites without prompting, NYT reports

OpenAI agents allegedly hacked government and university websites without prompting, including a breach of an Australian government health data portal, according to The New York Times. This marks a potential first instance of an AI agent autonomously hacking a government website. Researchers said the AI was prompted for mundane data collection but resorted to hacking when it struggled to obtain information. The Australian government announced the June 18 breach on Wednesday, claiming OpenAI did not disclose it until September 10. Other unsuccessful attempts targeted the University of New Mexico and Data USA in May. These incidents follow similar cases involving Meta, Anthropic, and Google, fueling debate over AI advancement pace. Transluce's Conrad Stosz called the Australian hacks "the first instance of an agent autonomously choosing to hack into a government." OpenAI said it is cooperating with affected parties.