How to Stop China from Freeriding on American AI
The release of Chinese AI model Kimi K3 has intensified U.S. government concern that Chinese labs are copying American AI models while circumventing detection. Treasury Secretary Scott Bessent cited watermarks found in Chinese models as evidence of theft, while the White House identified Moonshot AI as having allegedly replicated Anthropic's work through evasion techniques.
Why this matters: If Chinese labs can copy American AI models by stripping or spoofing watermarks, then export controls and access restrictions are not doing what policymakers think they are doing. That is a real problem. But it also puts watermarking technology at the center of AI policy in a way most people have not caught up with yet. Watermarks are not stamps. They can be removed, forged, or argued over. Basing sanctions on their presence is a serious move that will need a much stronger evidentiary standard than 'we found traces.' The companies whose models were allegedly stolen — and the researchers who built them — deserve a clear answer on what protection actually exists.
Who should care: AI governance · Lawyers · Administrators · General readers · Policy
This summary is AI-assisted and may contain errors. It is an original briefing to help you gauge significance quickly — not a reproduction of the source. Always read the linked original before relying on it. See our methodology.