The State Department is raising serious concerns about what it describes as widespread efforts by Chinese companies to siphon off American artificial intelligence technology—and yes, this is exactly the kind of thing Washington has been warning about for years.
According to a diplomatic cable sent to U.S. posts worldwide, officials have been instructed to alert foreign counterparts about the risks of “adversaries” extracting and replicating U.S. AI models. Among the companies named is DeepSeek, a Chinese startup that’s been making waves in the AI world—though now for reasons that go beyond innovation.
At the center of the issue is something called “distillation,” a process where smaller AI systems are trained using outputs from larger, more advanced models. Sounds efficient, right? Well, the concern is that these outputs may be coming straight from proprietary American systems—without permission, of course.
The warnings didn’t come out of nowhere. OpenAI previously told U.S. lawmakers that DeepSeek had been targeting American AI leaders in an effort to replicate their models. And now, the State Department is taking that concern global, making it clear this isn’t just a domestic issue—it’s a strategic one.
China, unsurprisingly, is rejecting the accusations outright. Its embassy in Washington called the claims “groundless” and accused the U.S. of trying to undermine China’s progress in the AI sector. Standard response—but one that doesn’t exactly calm concerns.
Meanwhile, DeepSeek is pushing forward, recently unveiling a preview of its new V4 model adapted for Huawei chips, highlighting China’s growing independence in AI development. Despite bans from several Western and Asian governments over data privacy concerns, its models remain widely used on global platforms. Because nothing says “no concerns here” like widespread restrictions paired with massive adoption.
The State Department cable also points to other Chinese firms, including Moonshot AI and MiniMax, as part of the broader concern. It warns that models created through these methods may appear competitive at a lower cost but often lack the full capabilities—and more importantly, the safeguards—of the original systems. According to the document, such models may even strip out security protocols designed to ensure neutrality and reliability.
All of this is unfolding just weeks before President Donald Trump is set to meet with Chinese President Xi Jinping in Beijing. So yes, the timing is about as convenient as you’d expect in an ongoing tech rivalry between two global powers.
Still, this moment underscores something bigger: protecting American innovation isn’t optional—it’s essential. And with stronger awareness and firm action, the U.S. is making it clear that it intends to stay ahead in the race that will define the future.