TY  - RPRT
TI  - AgentLSD: Evaluating AI Security Agents Under Adversarial Task Contamination
AU  - Matteo Golinelli
AU  - Idilio Drago
AU  - Matteo Boffa
AU  - Francesco Bergadano
AU  - Bruno Crispo
PY  - 2026
DO  - 10.1145/3847352.3848111
UR  - https://arxiv.org/abs/2609.19140
ID  - 2609.19140
ER  -