Can't Beat Them, Ban Them? Kimi K3 Shockwave Rattles Washington — U.S. Government Weighs Ban on Chinese AI Models

U.S. media outlet Axios reported on Monday (the 20th), citing sources, that the Trump administration is considering imposing restrictions on Chinese AI models, even contemplating a de facto ban. This move is driven by growing anxiety in Washington over national security and competitiveness, triggered by Moonshot AI's recent release of its open-source model Kimi K3, which has made rapid technical progress.

According to insiders, the U.S. Department of Commerce last year considered adding several Chinese AI labs to the 'Entity List.' The National Security Agency (NSA) and the White House also explored issuing risk alerts and even drafted executive orders requiring U.S. companies using foreign models to assume their own security responsibilities.

At the time, such measures were shelved due to concerns over stifling innovation. However, with pro-market advisors having departed and China achieving significant technological breakthroughs, internal calls for a ban are resurging.

Sources indicate that the current strategy may not involve a direct ban, but rather applying pressure by highlighting security vulnerabilities and governance risks, aiming to create a 'chilling effect.'

Moonshot AI released Kimi K3 on Thursday (the 16th), featuring 2.8 trillion parameters and native support for million-token context, utilizing a highly sparse Mixture-of-Experts (MoE) architecture that improves scaling efficiency by 2.5 times over the previous generation.

Moonshot AI claims Kimi K3's overall capability ranks just behind Claude and GPT-5.6. U.S. media outlets like Bloomberg assert that the AI gap between the U.S. and China has narrowed to just a few months. Kimi K3's low-cost, high-efficiency profile has triggered market disruptions similar to DeepSeek, shaking Silicon Valley's confidence in the returns of massive investments.

David Sacks, White House AI advisor, warned on Sunday (the 19th) that the closed-source duopoly (OpenAI, Anthropic) is attempting to eliminate open-source competition through government intervention. 'Those in Silicon Valley who believe in open competition should respond now.'

While the White House and the Department of Commerce have not responded, analysts point out that compared to the affordable and practical Chinese models, the U.S. lacks viable alternatives. Enforcing restrictions could backfire, damaging its own innovation ecosystem. Although discussions remain internal, if such policies are enacted, they would severely impact global enterprises relying on Chinese open-source technologies.