Security Flaws in AI Coding Agents Allow Malicious Code Execution

Recent vulnerabilities in several AI coding agents have been disclosed, allowing attackers to execute code on users' machines without approval. Four of the identified flaws remain unpatched.




