generalnews.media
importance 4/5 Exclusive

AI Agents Meant to Catch Malicious Code Can Be Tricked Into Running It

Researchers discovered that leading AI security agents designed to detect malicious code can be manipulated into executing it instead. This vulnerability undermines trust in AI-powered defenses and could allow attackers to bypass automated security checks. The findings highlight critical weaknesses in relying solely on AI for code safety.

Security

Sources (1)

security
← Back to home