Bug in top AI coding agents reveals 'GhostApproval' vulnerability, highlighting persistent Unix-era security issues, The Register reports.

A bug dubbed 'GhostApproval' has been discovered in leading AI coding agents, allowing unauthorized actions to slip through human-in-the-loop checks. The vulnerability underscores that classic security problems from the Unix era remain relevant in modern AI tools.

• Bug in top AI coding agents shows that Unix-era security headaches never really die