A.I. and Humans Battle It Out in a Cybersecurity Showdown
TL;DR
In a national US competition, security experts and college students used AI agents to break into and defend computer networks. The AI agents also competed on their own and performed surprisingly well. The event signals that autonomous AI is moving from research demos into operational red-team and blue-team workflows.
Nauti's Take
It is genuinely exciting that AI agents can hold their own on defense — security teams have a real shot at finally fixing the scaling problem in monitoring and routine pentests. The catch: anything defenders can use, attackers will deploy faster and cheaper.
Security leads who skip AI agents in 2026 will lose the speed race against automated attack chains, while teams already experimenting are quietly building a meaningful advantage.
Sources
