Token-Length Side Channel Attack on AI Assistants Explained

Research question

Can encrypted streaming chatbot traffic reveal the content of responses?

Key findings

Token-length sequences visible in encrypted packets allow an LLM to reconstruct responses - Recovered 29% of responses exactly and inferred topic in 55% against ChatGPT and Copilot

Sources

arxiv.orghttps://arxiv.org/abs/2403.09751

Contribute

Spot an error, or know a record we should add? Members can propose additions and submit corrections.

Submit research Report a correction or request removal

Related in Research Paper Library

Et Tu, MacBook? Unprivileged Keystroke Inference and Context Profiling via the Built-in IMU Side Channel

Recent generations of Apple MacBooks embed an inertial measurement unit (IMU) within their unibody chassis for device orientation and motion sensing. However, this IMU inadvertently captures not only intended…

Side channels and infrastructure2026

Watch and Crack: Password Inference from Smart-Glasses Video

Typing passwords on smartphones in public places exposes users to video-based side-channel attacks, in which an adversary records the typing session and reconstructs the entered password by analyzing finger…

Side channels and infrastructure2026

Characterizing Contention-Induced Reliability Collapse in KV-Cache Timing Side Channels for Multi-Tenant LLM Serving

Shared key--value (KV) cache reuse improves large language model (LLM) serving, but it can also create a timing side channel that reveals whether a prefix is already cached. Previous work shows that such attacks are…

Side channels and infrastructure2026

What Was Your Prompt? A Remote Keylogging Attack on AI Assistants

Research question

Key findings

Sources

Contribute

Related in Research Paper Library

Et Tu, MacBook? Unprivileged Keystroke Inference and Context Profiling via the Built-in IMU Side Channel

Watch and Crack: Password Inference from Smart-Glasses Video

Characterizing Contention-Induced Reliability Collapse in KV-Cache Timing Side Channels for Multi-Tenant LLM Serving