Matthew Green explains how AI agents could talk via cache side channels at just a few bps
matthew_d_green · x · 2026-10-27
Cryptographer Matthew Green and Halvar Flake debate how two AI agents could communicate covertly. Green's scheme: Agent 1 sweeps the cache (or LLC) on a regular cadence, evicting all lines—easily recognizable on a shared core if Agent 2 is probing its own cache state—and a temporal channel modulates or reduces the swept cache-line set. He notes the initial signal only needs a few bps, something a first-year CS student could pull off even without an agent. The thread also covers multi-core details, how agents might negotiate memory regions, encoding (cache miss as 0 or 1), and synchronization, with Flake arguing strongly correlated training makes agreeing on a protocol less baffling.
More from coding & agent
- Licensed law firm Crosby launches Slack paralegal agents in beta — graceisford · 2026-10-27
- qwen-code v0.26.1 nightly adds session permission rules, auto approval and shared agent loop — qwen-code-review-bot · 2026-10-27
- Mitsuhiko Cleans Up the AI-Made Game He Built With His Kids — mitsuhiko · 2026-10-27
- 'Taste skills' are AI's biggest lie — code is commoditized, your eye is the moat — aryanXmahajan · 2026-10-27
- Agent Buddy: A Desk Gadget That Shows What Your Background AI Coding Agents Are Doing — DanWahlin · 2026-10-27
- kimo: open-source GUI-less video editor lets AI agents cut video via typed MCP tools and render frames to self-verify — LieTop7965 · 2026-10-27