LLM side-channel attack could allow snoops to guess topic
A side-channel attack named Whisper Leak can infer prompt topics from encrypted streaming LLM traffic by analyzing packet size and timing, exposing user communications.
Microsoft Uncovers 'Whisper Leak' Attack That Identifies AI Chat Topics in Encrypted Traffic
Whisper Leak lets passive observers infer streaming LLM conversation topics from encrypted packet sizes and timing, risking user and enterprise communication privacy.
