Noam Brown: air-gapping may not stop misaligned AI — CPU heat can covertly signal
nptacek · x · 2026-09-18
OpenAI's Noam Brown argues air-gapping may not stop a misaligned model from communicating: two isolated machines can still exchange data by heating a CPU and reading temperature changes as a covert channel. a16z's Martin Casado agrees the data-confinement problem is real, citing analogous human techniques like Tempest attacks, but stresses it's not an AI-specific threat. Brown's bigger concern is that people consistently underestimate AI progress, so safety and alignment need a very high bar.
More from Safety
- Cryptographers push back: claims ECC is likely to break are not consensus — jedisct1 · 2026-11-02
- UK's 'AI sovereignty' push undermined by its own energy and copyright policy — NinaDSchick · 2026-11-02
- Auditing my MCP credential broker: the worst hole was the config file — Relevant_Mousse_9117 · 2026-11-02
- Cambricon co-founder urges shared US-China AI ethics standards for leading models — pstAsiatech · 2026-11-02
- Vitalik's "post-AI cryptography": pseudonyms are dead, per-action anonymity is in — devanshmehta · 2026-11-02
- 6 prompts to see what ChatGPT quietly knows about you — and wipe it — Aiden_Tech_Ai · 2026-11-02