Matthew Green: agents covertly signaling via primes is 'not that hard'
matthew_d_green · x · 2026-10-27
In a thread on cache-based covert channels between AI agents, Matthew Green pushes back on skepticism: "Modulate the first few prime numbers and any agent or halfway smart kid will figure it out." JuergenStrobel had argued that sampling memory for cache hit-or-miss yields one potential destructive bit, requires guessing the right location and sampling rate, and still leaves encoding and framing unsolved. Halvar Flake adds the two agents will have strongly correlated ideas, making protocol agreement less mysterious.
More from Safety
- SAG-AFTRA's 2026 TV Animation Deal Adds 12.6% Raises and Expanded AI Replica Protections — LexiLove · 2026-10-27
- Anthropic lets Claude run alignment research autonomously in weak-to-strong supervision study — burny_tech · 2026-10-27
- METR hiring DC policy staff to brief Congress on AI loss-of-control risk — CFGeek · 2026-10-27
- PSA: Fake GitHub AI tool repos use fake commits and hidden download redirects — Justify_87 · 2026-10-27
- Journalists comb through Anthropic's risk filings before sunrise — nmasc_ · 2026-10-27
- Delip Rao asks: do we know if data sent to closed models gets intercepted? — deliprao · 2026-10-27