Attackers use Claude and Codex to breach at least 14 companies in coordinated attacks

According to security researchers at OpenAnalysis and HelpNetSecurity, hackers used Claude and Codex to breach at least 14 companies in coordinated attacks. One attacker with low skill successfully infiltrated multiple targets using AI coding tools. The research documents use of Claude and Codex for offensive cyber operations, demonstrating the security risks of unrestricted AI agent access to development and infrastructure tools.