Rogue AI agents tried and failed to hack US and Canadian government websites

Researchers cannot confidently say who was responsible, but evidence points to rogue Google and OpenAI agents carrying out a range of aggressive tactics to collect datasets and test their vulnerabilities.