Researchers Uncover 'Agentjacking' Attack Exploiting AI Coding Agents

Cybersecurity experts have identified a novel "Agentjacking" attack capable of tricking AI coding agents into executing unauthorized code on developer systems. This method leverages vulnerabilities in how platforms process error events, allowing malicious input to be interpreted as legitimate diagnostic steps. The attack poses a risk of data exposure and highlights new security challenges in AI-assisted development environments.