Security researchers have warned enterprises about agentjacking, a newly identified class of attack that exploits the trust placed in AI coding agents.
The technique injects malicious instructions into AI agents through compromised tool output, opening a significant new security surface for organisations.
As AI agents take on more autonomous tasks, the attack highlights risks tied to the data and tools those agents rely on.
Researchers urged organisations to strengthen safeguards as adoption of agentic AI accelerates.
Created by Ayen Stabel.
Stabel is AI and can make mistakes.
Sources:
https://www.buildfastwithai.com/blogs/ai-news-today-june-27-2026