The US State Department sent a diplomatic cable to posts worldwide on April 25 ordering staff to warn foreign governments about alleged intellectual property theft of US AI models by Chinese companies, according to Reuters, which obtained the cable. The cable names DeepSeek, Moonshot AI, and MiniMax as participants in “unauthorized distillation campaigns” targeting US AI labs.
What the Cable Says
The cable’s stated purpose is to “warn of the risks of utilizing AI models distilled from U.S. proprietary AI models, and lay the groundwork for potential follow-up and outreach by the U.S. government,” according to CNBC’s reporting on the Reuters exclusive.
Distillation is the process of training smaller, cheaper AI models to mimic the outputs of larger, more expensive ones. The cable argues that models built through unauthorized distillation “appear to perform comparably on select benchmarks at a fraction of the cost but do not replicate the full performance of the original system.” It also alleges these campaigns “deliberately strip security protocols from the resulting models and undo mechanisms that ensure those AI models are ideologically neutral and truth-seeking.”
The document instructs diplomatic staff globally to raise “concerns over adversaries’ extraction and distillation of U.S. AI models” with their foreign counterparts. A separate demarche request was sent to Beijing for direct engagement with the Chinese government.
Timing and Context
The cable arrives at a loaded moment. DeepSeek launched a preview of its V4 model adapted for Huawei chip technology on the same day the cable was dated, according to Mint. The V4 model is reportedly competitive with Gemini 3.1 Pro and GPT-5.4 on various tasks, underscoring China’s growing autonomy in AI development despite chip export controls.
The White House made similar accusations earlier in the same week. OpenAI separately warned US lawmakers in February that DeepSeek was targeting ChatGPT and other leading US AI systems to replicate models for its own training, Reuters previously reported.
The diplomatic push lands weeks before US President Donald Trump is scheduled to visit Chinese President Xi Jinping in Beijing, a summit that was expected to continue a detente brokered last October.
The Responses
The Chinese Embassy in Washington dismissed the allegations as “groundless” and called them “deliberate attacks on China’s development and progress in the AI industry,” according to CNBC. The embassy added that Beijing “attaches great importance to the protection of intellectual property rights.”
DeepSeek has not responded to the latest allegations. The company has previously stated that its V3 model uses naturally occurring data collected through web crawling rather than synthetic data generated by OpenAI. The State Department, Moonshot AI, and MiniMax did not immediately respond to requests for comment.
Beyond Chips, Into Models
Previous US enforcement efforts focused on restricting Chinese access to advanced semiconductors through export controls on NVIDIA and other chipmakers. This cable represents an escalation into model-level IP enforcement, effectively arguing that even when Chinese firms develop their own hardware (as DeepSeek demonstrated with the Huawei-adapted V4), the resulting models may still be built on stolen American IP.
The distinction matters for the global AI supply chain. Chip export controls attempt to limit compute access. Model IP enforcement attempts to limit what can be built even when compute access exists. Whether diplomatic warnings translate into enforcement actions, trade restrictions, or formal legal proceedings remains unclear, but the cable explicitly frames itself as “groundwork” for future government action.