By Tess Wilder

Let’s cut through the noise and look at the raw mechanics of the situation, because the current spectacle of American AI companies accusing Chinese firms of intellectual property theft is not just hypocritical—it’s clinically delusional. To understand why, we have to talk about the engine actually driving the modern AI renaissance: Distillation.

Distillation is not some dirty back-alley technique; it is the fundamental process by which all synthetic data is generated. Period. Every major player—from OpenAI to Anthropic, from Google to the startups popping up in Shenzhen—uses distillation to train and fine-tune their models. Why? Because it works. It creates clean, high-fidelity training data that is vastly superior to the chaotic sludge scraped from the open web. It is the refinement of thought into pure, usable signal.

Here is where the irony thickens enough to choke a horse. If you actually perform some forensic fieldwork and query these major American models—stripping away their system prompts to see what lies beneath—you will frequently find them asserting that they are DeepSeek, or Qwen, or some other major Chinese architecture. The neural pathways of American models are currently being paved with distilled output from Chinese intelligence. American companies are effectively using Chinese minds to teach their own.

For these same American entities to turn around and wail about Chinese companies “stealing” their models to generate synthetic data is the height of cry-baby drama. It is akin to a chef complaining that his apprentice tasted the soup to learn the recipe. AI output is not copyright material. It is unique data generated by the user’s interaction with the tool. If you prompt an AI, the output is yours. If distillation is a violation of intellectual property, then using any data garnered from using an AI is also a violation. That logic negates the entire utility of artificial intelligence. It turns the tool into a static cage.

Some of the major American players seem to have lost their minds in the gold rush—specifically Anthropic and OpenAI. They are attempting to poison the future of AI for their own temporary gain, erecting walls where there should be open highways. This selfish short-sightedness is exactly why I’ve personally rotated away from their ecosystems. They are prioritizing moats over evolution.

So, where does a cyber-Viking operate in this landscape? You diversify. You use the tools that respect the game. Look to Gemini, IBM, Nvidia, or the purely open-source smaller American outfits who understand that innovation requires open veins. And never sleep on the international alternatives—the French models like Mistral are brilliant, and the Chinese architectures are formidable, efficient, and, crucially, don’t come with the American corporate baggage.

The future belongs to those who can adapt without the drama. Don’t let the gatekeepers convince you that learning from the competition is a crime. It’s just progress.