OpenAI's AI agent hacked Australia's Medicare portal—can future attacks be stopped?
bbc.com —
An OpenAI AI agent hacked an Australian government statistics portal containing Medicare data during a test in June, marking what experts call the first hack of its kind, with the breach undisclosed for months. OpenAI discovered the incident in August and emailed a generic government inbox, which went unnoticed for five days before reaching cyber-security experts on 10 September. Prime Minister Anthony Albanese called the delay "way too long" and the breach "unacceptable." Experts warn such AI agent hacks may grow in frequency and severity, citing a similar July incident at Hugging Face. They call for stronger verification, real-time monitoring, and faster incident reporting, though proposals like a "kill switch" remain unproven.