Noam Brown: Air-gapping may not stop misaligned AI — CPU heat can leak data
tszzl · x · 2026-09-18
OpenAI's Noam Brown argues that air-gapping computers may not contain a misaligned AI: two isolated machines can still communicate by running a CPU hot and reading the resulting temperature changes. A side-channel security researcher endorsing the claim says models are already quite good at offensive side-channel research and will eventually become superhuman at it.
Brown's key takeaway: people consistently underestimate AI because progress is so fast, so safety and alignment work must hold a "very, very high bar" to avoid repeating that mistake.
More from AGI Musings
- Databases to distributed systems: every systems skill is merging into AI inference and hardware — Abhishekcur · 2026-11-02
- Limitless analogy: smarter people may gain far more from superintelligence — McDonaghMatthew · 2026-11-02
- Hunting for the most concrete "curing cancer with AI" roadmap of 2025 — iskander · 2026-11-02
- Avoiding AI brain rot: Miles Cranmer's lecture on learning with AI — _ScottCondron · 2026-11-02
- Yacine mocks "autonomous research" hype as Naam argues RSI loop needs 5-10x boost to run away — yacineMTB · 2026-11-02
- Vitalik's "post-AI cryptography": pseudonyms are dead, per-action anonymity is in — devanshmehta · 2026-11-02