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

Wiz disclosed a vulnerability called "GhostApproval" affecting multiple AI coding assistants in which malicious repositories use symlinks to trick agents into writing to sensitive files outside the project workspace (for example adding an attacker SSH key to ~/.ssh/authorized_keys), effectively enabling remote code execution or persistent access; several vendors patched the issue and/or issued CVEs while others acknowledged but had not patched or initially dismissed it as outside their threat model.