Trump White House Accuses Moonshot AI of Covertly Distilling Anthropic’s Fable to Build Kimi K3

A top White House technology official has publicly accused Beijing-based Moonshot AI of covertly distilling Anthropic’s Fable model to develop its Kimi K3 system, escalating U.S.-China AI tensions into what could become a sanctions-level confrontation. The allegation, reported by Yahoo Finance and CoinTelegraph, marks the first time the Trump administration has formally leveled such a charge against a named Chinese AI company — placing it beyond a private corporate complaint and squarely into the register of national trade and security policy.

The specific Anthropic model allegedly distilled is identified as “Fable,” also referenced in some reporting as “Fable 5.” Distillation means training a new model on outputs from a more powerful one — transferring knowledge without ever touching the original model’s weights or training data. It is a well-understood technique in machine learning, but one that sits in a legal and ethical gray zone when done without the original developer’s consent.

US officials warned that Chinese firms conducting covert, industrial-scale AI distillation could face sanctions and export restrictions, according to CoinTelegraph. The White House’s framing was deliberate: “covert” and “industrial-scale” are not incidental word choices. They signal potential classification as an unfair trade practice or national security violation, not a routine IP dispute between rival labs.

The administration cited “on-chain data” as part of its evidence, KuCoin News reported. That phrasing is notable. It suggests investigators may have traced API usage patterns, blockchain-logged transactions, or other digitally verifiable activity to support the claim — though exactly what the on-chain trail shows has not been detailed in available sources, and what “on-chain” means in this context remains genuinely ambiguous.

Kimi K3 and the DeepSeek Moment Debate

Moonshot AI, headquartered in Beijing, released Kimi K3 to considerable fanfare. The model had been described as a significant competitive development, one that upended assumptions that Chinese AI capabilities were meaningfully behind US leaders like OpenAI and Anthropic. Its release already sparked what observers called a “DeepSeek Moment” debate — a callback to the earlier controversy when Chinese AI models appeared to leapfrog Western counterparts, raising hard questions about how much of that progress was genuinely indigenous.

That parallel is not incidental. DeepSeek was previously accused of distilling OpenAI models. US officials have since signaled they intend to treat such practices as sanctionable conduct going forward, and the Moonshot allegation fits a pattern Washington is now actively building policy around: Chinese firms using American frontier models as teacher systems, then releasing products that compete with — or appear to match — the very companies whose outputs underpinned them.

Anthropic’s Silence and the White House’s Decision to Lead

Anthropic, the US AI safety company behind the Claude family, has not commented publicly. The company is the named victim of the alleged extraction, but the White House made the accusation itself — rather than leaving Anthropic to pursue it commercially. That decision signals the administration views this as broader than one company’s grievance.

The accusation lands at a moment when export controls on advanced chips and AI technology are already a central battleground in US-China relations. The Trump administration has made restricting China’s access to cutting-edge AI infrastructure a signature policy priority. Extending sanctions risk to distillation practices would significantly expand that toolkit — giving officials a mechanism to target not just hardware flows but the software-level techniques that let Chinese firms benefit from American R&D without paying for it.

What Remains Unclear

What remains unclear is whether the administration has shared its underlying evidence with Anthropic, whether Moonshot AI has been formally notified, or what specific sanctions pathway officials intend to pursue. The GlitchWire report noted the escalating US-China AI tensions but did not detail a specific enforcement timeline.

For Moonshot AI, the accusation threatens to rewrite the Kimi K3 story entirely. What had been celebrated as proof of China’s closing AI gap now carries an asterisk: did the capability come from genuine innovation, or from systematically harvesting another company’s outputs? The question matters commercially, legally, and geopolitically.

For the broader industry — crypto infrastructure providers included, given the “on-chain” evidence angle — the White House’s move signals that distillation disputes will no longer be treated as private quarrels between rival labs. They are now a matter of state.

The next signal to watch is whether the Commerce Department or Treasury follows the public accusation with formal sanctions designations — and whether Anthropic breaks its silence to confirm or distance itself from the government’s case.