The Trump administration is not waiting for the next congressional hearing to escalate the AI cold war with China. It is using the full diplomatic apparatus of the United States government.

A diplomatic cable dated April 25 and obtained by Reuters instructs every American embassy and consulate on the planet to raise concerns with host governments about what Washington calls "widespread efforts by Chinese companies, including AI startup DeepSeek, to steal intellectual property from U.S. artificial intelligence labs."

The cable names three companies: DeepSeek, Moonshot AI, and MiniMax. It has not been previously reported.

The Mechanism: Distillation as Espionage

The specific accusation centers on distillation, the process of training smaller, cheaper AI models by feeding them the outputs of larger, more expensive ones. OpenAI warned U.S. lawmakers in February that DeepSeek was targeting ChatGPT and other leading American AI systems to replicate their capabilities for its own training data.

The cable goes further than previous White House statements. It claims these distillation campaigns "deliberately strip security protocols from the resulting models and undo mechanisms that ensure those AI models are ideologically neutral and truth-seeking."

In plain language: the State Department is alleging that Chinese firms are not just copying American AI. They are copying it and removing the safety guardrails.

The Timing Is Deliberate

This cable drops weeks before President Trump is scheduled to visit President Xi Jinping in Beijing. The diplomatic term for what the State Department is doing is a demarche: a formal step below a protest, above a polite request. The cable confirms a separate demarche has already been sent directly to the Chinese capital.

The timing is not accidental. Last October, Washington and Beijing brokered a detente on tech tensions. This cable tears that apart. Sending it globally, to every diplomatic post, signals that the administration wants allied governments to take sides before Trump sits down with Xi.

China Has Already Responded

The Chinese Embassy in Washington called the accusations "groundless" and "deliberate attacks on China's development and progress in the AI industry." Beijing maintains it "attaches great importance to the protection of intellectual property rights."

DeepSeek has previously stated that its V3 model used "naturally occurring" data collected through web crawling and did not intentionally use synthetic data generated by OpenAI. Neither Moonshot AI nor MiniMax responded to Reuters' request for comment.

The Subtext: DeepSeek V4 Launched the Same Day

On the same Friday the cable was sent, DeepSeek launched a preview of its highly anticipated V4 model, adapted for Huawei chip technology. The company that Washington accuses of stealing American AI is simultaneously proving it can run frontier-class models on entirely Chinese-made silicon.

Multiple Western and Asian governments have already banned their institutions from using DeepSeek over data privacy concerns. But the models remain among the most-downloaded on international open-source platforms. A global diplomatic warning campaign is not going to change that overnight.

What This Actually Means

The AI industry just crossed a threshold. Intellectual property disputes between tech companies are normal. Using the global diplomatic network to coordinate a warning campaign against specific foreign AI companies is unprecedented. This is trade war infrastructure applied to machine learning.

The cable explicitly states its purpose is to "lay the groundwork for potential follow-up and outreach by the U.S. government." Translation: this is the warning shot. Whatever comes next will be louder.

Original reporting by Raphael Satter, Reuters.