Dr. Ben Nassi

My research focuses on the security and privacy of systems.

AI Security. In the last three years, I have mostly investigated the security and privacy of LLM-powered applications (such as Gemini and Copilot) and AI-powered systems (such as Tesla).

Side-channel Attacks. I demonstrated a few methods to eavesdrop speech and recover secret keys (cryptanalysis) using optical sensors (video cameras and photodiodes).

Cyber-physical Systems. I researched privacy aspects of drones and developed privacy enhancing methods to detect illicit drone video streaming.

AI-powered System Security

Invitation Is All You Need! Promptware Attacks Against LLM-Powered Assistants in Production Are Practical and Dangerous (BHUSA'25) - webpage

Here Comes the AI Worm: Preventing the Propagation of Adversarial Self-Replicating Prompts Within GenAI Ecosystems (CCS'25) - webpage

Phantom of the ADAS: Securing advanced driver-assistance systems from split-second phantom attacks (CCS'20) - webpage

Side Channel Attacks (Eavesdropping, Cryptanalysis)

Video-based Cryptanalysis (S&P'24) - webpage

Optical Cryptanalysis (CCS'23)

Lamphone (USENIX Sec'22) - webpage

Glowworm Attack (CCS'21) - webpage

Cyber-Physical Systems Security

 SoK - Security and Privacy in the Age of Commercial Drones (S&P'21)

Drones Cryptanalysis (S&P'19) - webpage