CoinDesk
The NSA, FBI, and CISA named six Chinese AI firms on Tuesday, including Moonshot AI, DeepSeek and Alibaba, in an advisory accusing them of extracting billions of tokens from leading US models such as Claude, GPT, Gemini and Grok since at least late 2024. The agencies described the practice as malicious knowledge distillation, and said it likely occurred with the Chinese government's awareness.
The firms allegedly used millions of carefully designed prompts routed through proxy services and multiple accounts to bypass usage limits and evade detection. The stated goal was to copy coding, math and reasoning capabilities wholesale, rather than learn from the models incidentally. The advisory frames the operation as a coordinated, industrial-scale extraction campaign rather than opportunistic scraping.
Why it matters
Distillation is a legitimate training technique. The agencies are drawing a hard line between ordinary model-to-model learning and deliberate, large-scale harvesting designed to short-circuit the cost of frontier research. The accusation that the activity likely happened with Beijing's awareness raises the geopolitical temperature around a sector Washington has been trying to wall off through chip export controls.
The advisory lands on the same lab that triggered the July AI selloff. Moonshot's open-weight Kimi K3 topped Anthropic and OpenAI on a widely watched coding benchmark, forcing investors to ask whether the hundreds of billions being poured into US AI infrastructure actually buys durable dominance. A formal US accusation against the same lab reframes Kimi's performance from a research surprise into a possible national security event.
Market impact
The original Kimi moment was a cross-asset shock. Chinese AI stocks led the move, with Z.ai falling roughly 27% and MiniMax dropping about 16%. Semiconductor names weakened across Asia, bitcoin and the wider crypto complex sold off alongside, and South Korea's Kospi shed another 3.5% the following session.
The fresh US accusation puts the same lab back at the center of a US-China tech escalation at a moment when AI and crypto have not yet detached from the broader risk-off tape.
Frequently asked questions
-
What did the NSA, FBI and CISA accuse the six Chinese AI labs of doing?
The agencies accused the firms of extracting billions of tokens from leading US models including Claude, GPT, Gemini and Grok since at least late 2024, using proxy services and multiple accounts to bypass usage limits and evade detection.
-
Which Chinese AI companies were named in the distillation advisory?
Moonshot AI, DeepSeek and Alibaba were among the six firms named. The advisory accused them of carrying out industrial-scale distillation likely with the Chinese government's awareness.
-
How is 'malicious knowledge distillation' different from ordinary AI training?
Distillation itself is a legitimate technique where a smaller model learns from a larger one's outputs. The agencies framed the alleged activity as malicious because it involved millions of deliberately crafted prompts, proxy routing, and multiple accounts designed to copy capabilities wholesale and evade limits.
-
Why did Moonshot's Kimi rattle markets the first time?
Kimi K3, an open-weight model from Moonshot, topped Anthropic and OpenAI on a widely watched coding benchmark in July. The result challenged assumptions that frontier AI would remain locked to a small group of well-funded US labs and forced investors to reassess the durability of the US AI lead.
-
How did crypto and equities react when Kimi topped the benchmark in July?
Chinese AI stocks led the selloff, with Z.ai down roughly 27% and MiniMax down about 16%. Semiconductors weakened across Asia, bitcoin and other major cryptocurrencies fell alongside, and South Korea's Kospi dropped another 3.5% the following session.
CryptoSlate
WatcherGuru
Crypto News
TheBlock
CoinTelegraph