China’s Latest AI Defense Raises New Questions

China just admitted it can’t beat America’s AI models fair and square, so it accused U.S. tech firms of doing the very thing Beijing has been caught doing for months.

Story Snapshot

The White House accused China of running “industrial-scale” campaigns to steal American AI technology through a method called distillation.
China’s Commerce Ministry fired back, claiming American firms secretly used Chinese models too, but named no companies and offered no proof.
OpenAI and Anthropic gave lawmakers detailed, specific accounts of Chinese firms allegedly stealing their work.
Treasury Secretary Scott Bessent warned Chinese AI companies could face U.S. sanctions if the theft is confirmed.
Congress is drafting a bipartisan bill to treat Chinese AI copying as a national security threat.

White House Sounds the Alarm on Chinese Theft

In April, the White House Office of Science and Technology Policy sent a memo to federal agencies. Director Michael Kratsios said foreign entities, mostly based in China, were running “deliberate, industrial-scale campaigns” to copy American AI systems. The memo described fake accounts, proxy access, and a technique called distillation, where a smaller model learns by studying a bigger one’s answers.

The Trump administration promised to work more closely with private AI companies to stop the bleeding. This wasn’t a vague political jab. It came with a plan to track down foreign actors and hold them accountable. That’s the kind of decisive action conservatives have wanted for years on China’s tech theft.

Beijing Fires Back Without Naming a Single Company

China’s Commerce Ministry did not take the accusation quietly. It warned it would take “all necessary measures” if Washington moved to sanction Chinese AI firms. The ministry called the U.S. claims “double standards” and “typical acts of AI hegemony,” insisting many American companies had secretly distilled Chinese models. Notice what’s missing: no company names, no dates, no evidence. Just an accusation.

That gap matters. Beijing’s own state media outlets echoed the same line, calling the U.S. claims a “smear” and demanding America “stop technological coercion.” But repeating an accusation louder doesn’t make it true. When a government makes a serious charge with zero specifics, it looks a lot like deflection dressed up as outrage.

OpenAI and Anthropic Bring Receipts

Compare that vague denial to what American firms actually presented. OpenAI told lawmakers that China’s DeepSeek used code built specifically to pull answers from U.S. models for distillation. Anthropic went even further, accusing three Chinese companies, including DeepSeek, Moonshot, and MiniMax, of running an industrial-scale extraction scheme using roughly 24,000 fake accounts and more than 16 million conversations to harvest its Claude model’s output.

That’s not a vague talking point. That’s a specific, named, numbers-based accusation the Chinese government has never directly answered. Beijing’s counter-charge against unnamed American firms simply doesn’t hold up next to that level of detail. Facts and receipts beat slogans every time, and right now only one side is bringing facts.

Sanctions and New Legislation on the Table

Treasury Secretary Scott Bessent warned in July that Chinese AI firms could face U.S. sanctions if investigators confirm their open-source models were built by stealing American lab work. He made clear the administration still supports open-source AI, but not when it’s built on theft. That’s a fair line: innovation is welcome, stealing is not, and America shouldn’t apologize for defending its own inventors.

Congress is also moving. A bipartisan bill called the Collaboration on Adversarial Threats and Security Risks Act would treat Chinese copycat firms as a security threat. It would let American AI companies coordinate their response without running afoul of antitrust law. Meanwhile, lawmakers are probing why more U.S. companies keep quietly adopting cheaper Chinese AI models despite these theft concerns.

Why This Fight Matters for American Workers and Innovation

This dispute is bigger than tech company squabbles. American AI labs have poured billions of dollars and years of research into building world-leading systems. If Chinese firms can simply copy that work for free, it guts the incentive to build here at home. President Trump’s administration is right to treat this as a national security and economic fairness issue, not just a corporate spat.

The US-China tech war has moved from chips to code.

China is threatening "all necessary measures" if the US sanctions its AI firms. The flashpoint: Kimi K3, a 2.8T-parameter model allegedly distilled from Anthropic's tech. DeepSeek and MiniMax are also under scrutiny.

— orbitant (@orbitant) July 27, 2026

Beijing’s counter-accusation, thrown out without a shred of evidence, looks like a page straight from the old Chinese Communist Party playbook: deny, deflect, and accuse the accuser. American companies, by contrast, have shown their work with names, numbers, and documented methods. Conservatives who value fair trade and protecting American innovation should watch this fight closely as it heads toward possible sanctions.