6 papers
MLCommons Jailbreak Benchmark v1.0
Carsten Maple, Cagatay Yucel, Isaac Holeman +33
Modern AI systems are designed to refuse hazardous requests. A jailbreak is a prompt crafted to bypass those safeguards and elicit outputs that the system would normally refuse to…
Multi-Task Consistency-based Detection of Adversarial Attacks
Cong Chen, Jean-Philippe Monteuuis, Jonathan Petit
Deep Neural Networks (DNNs) have found successful deployment in numerous vision perception systems. However, their susceptibility to adversarial attacks has prompted concerns regar…
The Great Pretender: A Stochasticity Problem in LLM Jailbreak
Jean-Philippe Monteuuis, Cong Chen, Jonathan Petit
"Oh-Oh, yes, I'm the great pretender. Pretending that I'm doing well. My need is such, I pretend too much..." summarizes the state in the area of jailbreak creation and evaluation.…
Uncertainty Quantification for Collaborative Object Detection Under Adversarial Attacks
Huiqun Huang, Cong Chen, Jean-Philippe Monteuuis +2
Collaborative Object Detection (COD) and collaborative perception can integrate data or features from various entities, and improve object detection accuracy compared with individu…
V2X Misbehavior in Maneuver Sharing and Coordination Service: Considerations for Standardization
Jean-Philippe Monteuuis, Jonathan Petit, Mohammad Raashid Ansari +2
Connected and Automated Vehicles (CAV) use sensors and wireless communication to improve road safety and efficiency. However, attackers may target Vehicle-to-Everything (V2X) commu…
V2X Misbehavior and Collective Perception Service: Considerations for Standardization
Mohammad Raashid Ansari, Jean-Philippe Monteuuis, Jonathan Petit +1
Connected and Automated Vehicles use sensors and wireless communication to improve road safety and efficiency. However, attackers may target Vehicle-to-Everything communication. In…