Identity-Based AI Attack Threatens Security of Enterprise Data
Researchers at Noma Labs have identified a new AI attack vector called "workflow identity hijacking" that bypasses security controls by exploiting a design flaw in enterprise AI pipelines. Attackers can send seemingly innocuous requests through unauthenticated entry points, such as support emails or GitHub issues, to trigger the AI workflow to retrieve and deliver information belonging to another user, specifically a high-privilege user like a finance director. This attack isn't about tricking an AI model, but rather about leveraging the AI workflow's trust in AI-generated output and the lack of proper identity verification. Mitigation strategies involve shifting security controls to application and infrastructure layers, implementing identity-aware token delegation, and establishing contextual authorization checkpoints.
Summary written automatically in our own words from the original article, which belongs to its publisher and remains the reference. It may contain errors. Sources & data
_ber1a_Alamy.png?width=720&quality=80&disable=upscale)