Reported AI-Agent-Assisted Cyber Operations Put Autonomy Claims to the Test
A report describes one operator coordinating 12 AI agents across 11 countries, with reported confirmed data theft from emergency services, universities, and financial targets. Reported refusals, hallucinations, and false positives complicate claims of reliable autonomous attacks; independent validation and concrete defensive lessons remain key.
Sources (2)
Updated Nov 19, 2026