America Just Handed China Two Weeks — And China Took Them

Anthropic is back online after its export-control shutdown. But developers have already fled to DeepSeek, and even Trump's former AI czar admits: the crackdown was a gift to Beijing

The Partial Return — And the Damage Already Done

After a two-week blackout that watchchina.ai has tracked since it began, Anthropic's most advanced AI is cautiously coming back online. After a two-week shutdown due to an export control directive, Anthropic was allowed by the White House on Friday to release its powerful Mythos 5 model to some companies and federal agencies, though its Fable 5 model remains off the market. The restoration is partial, conditional, and — critically — too late to undo the strategic consequences already set in motion.
The framing from within America's own AI establishment is remarkably blunt. The Trump administration's crackdown on Anthropic's leading artificial intelligence models is looking like a gift to the country's top adversary in the AI race: China. This is not the assessment of a Beijing spokesperson or a China-friendly commentator. It is the plain reading of events by CNBC's business desk — and, as it turns out, by some of the most hawkish figures in the American AI policy world.

When the Hawks Start Warning About Their Own Policy

The most striking signal came from a man who has been one of Anthropic's fiercest critics. Even former Trump crypto and AI czar David Sacks, who has been a vocal critic of Anthropic's approach to AI safety, wrote a cryptic post on X above a screenshot of a Wall Street Journal headline stating that China has matched Anthropic in cybersecurity. "A year ago, President Trump declared that America was in a global AI race and that the way to win it was to be pro-innovation, pro-infrastructure, pro-energy, and pro-export," Sacks wrote. "President Trump was exactly right; we deviate from that strategy at our peril."

When the administration's own former AI czar publicly warns that the current strategy is a deviation from the winning formula — using a headline about China matching American cybersecurity capabilities as his evidence — the internal contradiction at the heart of US AI policy becomes impossible to ignore. The timing could hardly be worse. The shutdown lands just as much of corporate America is moving from an era of so-called tokenmaxxing — allowing developers to spend on AI without restraint — to a focus on efficiency and return on investment. That also plays into China's hands.

The Developers Have Already Voted — With Their Traffic

The abstract strategic argument becomes concrete in the decisions of individual companies, and those decisions are flowing in one direction. Earlier this month, Flo Crivello, CEO of AI startup Lindy, switched his company off Anthropic's Claude models and moved 100% of its traffic to DeepSeek. "We did it, and you could see that cost curve go down, like, crash to the ground," Crivello said.

The structural reason this migration is so difficult for the US to stop is one watchchina.ai has emphasised repeatedly: openness is a strategy that export controls cannot touch. Chinese AI developers are reaching US users with ease because of how simple it is for a company to download open-weight models and run them on their own servers without relying on a third-party cloud. There is no API to cut off, no access to revoke, no nationality to verify. And the capability gap that once justified paying a premium for American models is closing precisely where it matters most to security-conscious buyers. Travis Lanham, co-founder of AI security startup Armadin, which is experimenting with GLM-5.2 and Kimi K2.7 from Moonshot AI, said the models are showing improved capabilities for cybersecurity use cases like analyzing reconnaissance data and creating exploit code — describing the open-weight model landscape as "kind of the Wild West."

The verdict of the past two weeks writes itself. America restricted its best AI over a China suspicion. China released its best AI for free. American developers migrated. And when the restriction was partially lifted, the developers did not come back — because on the other side, the models were cheaper, open, and now good enough even for the cybersecurity work that was supposed to be America's last redoubt.

Two weeks was all it took. China did not waste them.

Sources: CNBC, Wall Street Journal, The Washington Post