Blue Teaming Function-Calling Agents
Abstract
We present an experimental evaluation that assesses the robustness of four open source LLMs claiming function-calling capabilities against three different attacks, and we measure the effectiveness of eight different defences. Our results show how these models are not safe by default, and how the defences are not yet employable in real-world scenarios.
Fields of Study & Subjects
Adversarial Robustness in Machine Learning
Security and Verification in Computing
Cryptographic Implementations and Security
Publication details
- Journal / source
- arXiv (Cornell University)
- Publication date
- 14 January 2026
- Document type
- Preprint