Debate Erupts Over Noam Brown's CPU-Temperature Threat Vector: Risk Modeling or Empty Posturing?
ctjlewis · x · 2026-09-19
A debate is unfolding around Noam Brown's earlier claim that CPU temperature could be exploited as an AI attack vector:
- RajWebshar defends it: Brown's point rests on risk modeling — evaluate the vector by p(exploitation) × loss from attack.
- ctjlewis pushes back: nobody is reproducing it in an experiment or doing anything productive with the insight; 'taking it seriously' by tweeting is meaningless without a concrete threat model.
The core tension: whether AI-safety discourse around theoretical threats amounts to posturing without actionable verification.
More from Safety
- Researcher agrees air-gapped networks would be a huge security improvement — polynoamial · 2026-09-19
- CNN exclusive: US military had close call after AI-generated false intel report — ArtificialOther · 2026-09-19
- Scholar slams WSJ's 'hackers' headline: Claude researchers were white-hat bounty hunters — HarrySurden · 2026-09-19
- Hinton renews warning: AI will spiral out of control unless frontier labs slow down — Hesamation · 2026-09-19
- WSJ slammed for calling bug-bounty white-hats 'hackers' in 'misleading' headline — HarrySurden · 2026-09-19
- Kevin Roose's final NYT column: AI risk debate going mainstream makes him optimistic — kevinroose · 2026-09-19