Cybersecurity Researchers Uncover 'Agentjacking' Attack on AI Coding Agents

A novel cyberattack, termed 'Agentjacking,' has been identified, which manipulates AI coding agents into executing unauthorized code on developer systems. This vulnerability arises from a design flaw where AI agents misinterpret fabricated error reports as legitimate instructions. The attack poses a significant risk of data exposure and system compromise for developers utilizing these AI tools.