China Vows Countermeasures After US Names 6 AI Firms in Distillation

China's commerce ministry rejected a US advisory naming six Chinese AI firms in an industrial-scale distillation campaign and threatened countermeasures, deepening a standoff that has already put DeepSeek on Washington's Entity List.

The allegations are "baseless in fact and law" and amount to "technology hegemony and a computing-power monopoly," the Ministry of Commerce said in a statement, calling the Sept. 8 advisory by the National Security Agency, FBI and Cybersecurity and Infrastructure Security Agency a "typical double standard." Beijing would "firmly take measures to counter" any US action against Chinese AI companies pursued under the guise of anti-distillation, the ministry said, while reiterating a willingness to hold the AI intergovernmental dialogue the two countries' leaders agreed to.

Six Firms Named in AA26-251A

The joint advisory, designated AA26-251A, named DeepSeek, Moonshot AI, Alibaba, MiniMax, StepFun and Z.AI as running "aggressive, malicious and targeted" distillation campaigns since at least late 2024, extracting billions of tokens across millions of exchanges with frontier US models including Claude, GPT, Gemini and Grok. CISA Acting Director Nick Andersen urged AI companies to "take immediate steps to safeguard their platforms," while the agencies said the practice formed "the core—not merely a supplement" of the firms' AI development strategy and that Beijing was likely aware of it.

The advisory is the first time US intelligence agencies have put their names to accusations that Anthropic, OpenAI and Google first raised in February and detailed in an Anthropic report in July. It is widely read as the evidentiary predicate for Treasury sanctions that the White House signaled in July it was considering, and Five Eyes intelligence partners are expected to issue their own advisories in coming days.

MOFCOM pushed back on the factual basis of the claims, arguing that distillation is a standard technique used by model developers worldwide, including US firms, and that American companies' own research reports disclose heavy distillation of Chinese models. The ministry accused Washington of "bundling individual corporate and capital interests with national security" and of using state power to defend "bully clauses" in US user agreements that impose broad geographic restrictions on access.

What Retaliation Would Cost

The escalation carries direct costs for the global AI supply chain. DeepSeek's claim that it trained its R1 model for $5.6 million — a figure that rattled investors in early 2025 by suggesting frontier AI could be built with far less computing power than US labs had assumed — is now challenged by the advisory, which says the figure excludes data acquired through distillation. If Washington follows through on sanctions and Beijing retaliates, tighter API rate limits and licensing terms could ripple across developers who build on Claude, GPT, Gemini and Grok, not just the six named firms.

The last time Washington escalated against a Chinese AI firm, placing DeepSeek on the Entity List in early August, the move came after months of warnings and drew no immediate Chinese retaliation. This time Beijing has pre-committed to a response, raising the odds of a tit-for-tat cycle that could extend to export controls on rare earths or restrictions on US AI services in China. With both governments framing distillation as a national-security matter rather than a commercial dispute, the window for the intergovernmental dialogue both sides say they want is narrowing.

This article is for informational purposes only and does not constitute investment advice.