US AI Rules Miss Software Gap as China Extracts Models
SkimNews Take
Export controls were architected around physical goods, so the deeper issue isn't enforcement gaps but a category error: AI capability now travels as copyable code, meaning bilateral diplomacy can only patch a framework that still treats software as an adjunct to hardware.
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- U.S. hardware export controls target physical goods like lithography equipment and GPUs, but they do not cover AI software, leaving a policy gap in China‑U.S. tech diplomacy.
- Anthropic disclosed that Chinese labs DeepSeek, Moonshot AI, and MiniMax conducted large‑scale model extraction campaigns using tens of thousands of fraudulent accounts.
- OpenAI filed a parallel report with the House Select Committee on the Chinese Communist Party detailing similar AI extraction concerns.
- White House Office of Science and Technology Policy issued a memorandum accusing China‑based entities of industrial‑scale campaigns against U.S. frontier AI systems.
- President Trump is scheduled to visit Beijing on May 14, providing an early test of whether the two governments will address the software‑level AI gap.
Why it matters: U.S. AI firms risk losing their edge as Chinese labs use tens of thousands of fake accounts to steal model capabilities, forcing a policy scramble ahead of the May 14 Trump‑Beijing summit.
