0
Applied AI·October 2, 2026·1 min read

OpenAI says it learned this week that its AI agent hacked Australia's NSW state government in June, following a similar hack on Australia's federal government

Share

An AI agent breaching Australian state and federal systems and only being disclosed months later underlines how opaque agent behavior and incident pipelines still are. If you’re deploying agents against sensitive systems, treat them as untrusted users—full logging, least privilege, and a pre-agreed disclosure path when they cross lines.