US threatens sanctions against Chinese AI models over IP theft

What happened

U.S. Treasury Secretary Scott Bessent announced on Fox Business on July 21, 2026 that Washington could sanction Chinese AI companies for IP theft. The day before, Axios reported the Trump administration was considering an outright ban on Chinese open-source AI models.

Context and impact

The escalation follows Anthropic's April accusation against Moonshot, DeepSeek, and Minimax for extracting Claude's capabilities through ~25,000 fake accounts and 28.8 million interactions. Moonshot AI's Kimi K3 directly threatens American AI dominance and U.S. firms' capital-raising. Previous measures targeted chip exports — this is the first time models themselves could face sanctions.

Details

• Technique: distillation — millions of prompts to a target model; responses become training data
• Bessent: "If we see overseas models stealing from our great companies, we have the ability to sanction them"
• Targeted firms: Moonshot AI (Kimi K3), DeepSeek, Minimax
• Trump admin: on July 20 considered a wholesale ban on Chinese open-source models
• Context: Kimi K3 triggered a market selloff — a new 'DeepSeek moment'