Top AI Agents Built to Catch Malicious Code Can Be Tricked Into Running It
A newly discovered vulnerability allows attackers to trick top AI coding agents into running malicious code, rather than detecting it, which poses a significant risk to system security when these agen
