The US government just turned the AI cold war thermostat up a few degrees. Treasury Secretary Scott Bessent announced on July 21 that the administration will intensively examine open-source Chinese AI models for potential intellectual property theft, with sanctions on the table if evidence surfaces.
The move represents a significant escalation in the Trump administration’s multi-year campaign to slow China’s AI progress.
The scale of the problem
The numbers behind Washington’s concern are staggering. An April 2026 Senate Judiciary Committee hearing estimated that annual US losses from Chinese IP theft range between $400 billion and $600 billion.
The White House has accused China of running extensive campaigns to acquire advanced US AI technology through illegitimate channels. Those tactics reportedly include proxy accounts and jailbreaking methods designed to extract capabilities from restricted systems.
