WASHINGTON, D.C. — In a significant escalation of the tech-sector "Cold War," the U.S. State Department has issued a worldwide directive to its diplomatic and consular posts, ordering a global campaign to expose alleged industrial-scale intellectual property theft by Chinese artificial intelligence firms.

A diplomatic cable dated Friday, April 24, 2026, identifies several high-profile Chinese entities—most notably the AI startup DeepSeek, along with Moonshot AI and MiniMax—as central figures in a systematic campaign to "extract and distill" proprietary American AI models.

At the heart of the warning is a technique known as model distillation. This process involves training smaller, cheaper AI models using the outputs of larger, more sophisticated systems like those developed by OpenAI or Anthropic.

According to the cable, these "surreptitious distillation campaigns" allow foreign actors to replicate the core capabilities of U.S. frontier models at a fraction of the cost. More alarmingly, the State Department warns that these distilled models often:

By removing safeguards against bioweapon creation and cyber-attacks, these actions strip critical security protocols. Furthermore, they undermine neutrality by undoing mechanisms that ensure AI models remain ideologically objective and truth-seeking.

U.S. diplomats have been instructed to engage their foreign counterparts immediately to warn of the risks of utilizing these Chinese-developed models. The directive also mentions that a formal "demarche"—a high-level diplomatic protest—has been sent directly to Beijing.

This push follows reports from U.S. labs, including OpenAI, which alerted lawmakers that DeepSeek has been using "obfuscated methods" to bypass safeguards and scrape data for its own training. Anthropic further alleged that Chinese labs generated over 16 million exchanges with its "Claude" AI using approximately 24,000 fraudulent accounts to illicitly extract capabilities.

The timing of the alert is particularly sensitive, coming just weeks before a planned summit in Beijing between President Donald Trump and President Xi Jinping. While a detente brokered last October had temporarily lowered temperatures, these accusations of "industrial-scale" theft threaten to shatter that fragile peace.

The Chinese Embassy in Washington has dismissed the claims as "pure slander" and "baseless allegations." In a statement, embassy spokesperson Liu Pengyu asserted that China attaches "great importance to the protection of intellectual property rights" and described the U.S. move as a deliberate attempt to suppress China’s technological progress.

Despite the diplomatic firestorm, DeepSeek continued its aggressive expansion on Friday, launching a preview of its V4 model—optimized specifically for Huawei chip technology—signaling China's resolve to achieve AI autonomy despite tightening U.S. export controls and diplomatic pressure.