A newly identified attack category that injects malicious instructions into AI agents via compromised tool output is creating a significant security surface for organisations.
Tag: agentjacking
AI Coding Governance Gap Grows as 97% Adoption Meets Only 33% Internal Governance — Agentjacking Risk
A major survey of enterprise software teams reveals that while nearly all developers use AI coding tools fewer than a third have governance policies — creating security vulnerabilities.
Agentjacking Attack Hits 2388 Organisations With 85 Percent Exploitation Success Rate
The newly identified attack class compromises AI coding agents exploiting trust relationships between agents and underlying systems to steal credentials and data.