Bessent Blames Chinese AI Models for Attacks, Floats Sanctions
Treasury Secretary Scott Bessent says Chinese AI models are behind AI-powered attacks, pointing to a hack that knocked out refrigeration in French grocery stores. He's also raising the threat of sanctions and Entity List designations for models he claims were built on American research. None of his attributions have been independently verified.
- The source published it on Nov 20, 2026 at 3:27 PM.
- We couldn't link the original source.
Speaking on November 12–13, Bessent said a Chinese AI model was used to disable refrigeration systems in roughly 150 to 700 French grocery stores. He pinned the operation on Iranian operatives who were allegedly wielding the Chinese model — AI as the tool, Iran as the hand holding it. The figure's range is wide, with the top end more than four times the bottom, and no independent verification has confirmed the Chinese-model link.
The escalation didn't start there. On July 21–22, Bessent said Washington would scrutinize Chinese AI models for potential intellectual property theft, zeroing in on what he called “industrial-scale distillation” — training a cheaper model on a stronger one's outputs without permission. He floated the idea that models like Moonshot AI's Kimi K3 could carry “watermarks” exposing training on US model outputs, and said sanctions or Entity List placement were possible consequences. Those were possibilities he raised, not actions taken.
In September, US agencies accused six Chinese firms of running distillation attacks on American models dating back to late 2024. Kimi K3 was singled out in connection with large-scale distillation from Anthropic's Fable model. The same month, US officials proposed a hotline with China to manage AI risks, including runaway agents and cyberattacks.
The concrete fallout would land on companies tied to Chinese AI developers. If scrutiny turns into sanctions or Entity List designations, firms like Moonshot AI could struggle to work with American partners, suppliers and customers. US developers face pressure too: treating distillation as a national security matter gives firms like Anthropic stronger incentives to lock down how their outputs get harvested at scale.
The watermark idea matters because it could hand regulators a technical basis for action instead of a purely political one — if US officials can credibly show a Chinese model carries traces of American outputs. Attribution is the other open question, since Bessent's French grocery store account remains unconfirmed and the store count still spans a wide range. Next up: whether Treasury or other agencies formally move against Moonshot AI or the six firms named in September, whether independent researchers or French authorities weigh in on the refrigeration attack, and whether the proposed US-China AI hotline survives the increasingly adversarial tone.
TopicsScott Bessent · Moonshot AI · Kimi K3 · Anthropic · Fable · US Treasury · Entity List · China
Written by AI from verified reporting.
More stories
- Tech & AI cryptobriefing.com 0 viewsHouse committee moves to boot Chinese open-weight AI from US government systems
- Tech & AI thedefiant.io 2 viewsIntel 18A vs TSMC N2: Claimed Gains Aren't a Chip Race
- Tech & AI beincrypto.com 2 viewsBridgewater Data Shows Cyber Attacks Doubled in Just Three Months
- Tech & AI finance.yahoo.com 2 viewsAnthropic's Blowout Debut Costs Rival AI Stocks $495B in a Day

