About

Who we are

The Agentic AI Security Initiative is an effort part of the NDS2 Lab focusing on the security of agentic AI systems and the use of agentic AI systems to improve systems security.

What we work on

Our work spans the full stack of agentic AI: from the infrastructure that mediates what autonomous agents can do, to the adversarial attacks that target them, to the defenses that contain their failure modes. Results are published at top-tier security and machine learning venues, released as open-source tooling, and developed in close collaboration with industry and government partners.

Collaboration

We welcome inquiries from prospective students, collaborators, and sponsors. The People page lists current members and their websites; reach out directly to discuss research interests or partnership opportunities.