More than 20 companies, including tech giants Nvidia, Microsoft, and Meta, urged policymakers on Friday to avoid "premature restrictions" on open-weight artificial intelligence models, even as U.S. officials accused Chinese firms of stealing American intellectual property in the same sector. These transnational corporate interests claim that unrestricted access to AI models strengthens competition and ensures the technology's benefits are "broadly shared rather than concentrated in a few hands." Their letter warned that "relying solely on closed models is not inherently safe," arguing that concentrating advanced AI capabilities behind a small number of closed models compounds risk.
This corporate push for open access comes as Chinese open-weight models, notably Moonshot AI's Kimi K3, have gained traction, outperforming some cutting-edge American offerings across industry benchmarks. U.S. Treasury Secretary Scott Bessent stated on Tuesday that the Trump administration would investigate whether Chinese companies were stealing American intellectual property. He affirmed the government's "ability to sanction them because of this theft."
White House advisor Michael Kratsios added on Wednesday that China's Moonshot AI developed Kimi K3 by distilling Anthropic's technology. While acknowledging that legitimate AI distillation plays a vital role in the open innovation ecosystem, Kratsios warned that "large-scale, covert industrial distillation aimed at stealing proprietary U.S. technology" is "unacceptable." This distinction draws a new line on China, with the administration backing open-weight models while simultaneously accusing Chinese companies of theft.
Elite Interests and National Security
Axios reported that the administration is effectively defining two categories of distillation: small-scale or authorized distillation, treated as normal development, and industrial-scale distillation conducted through fraudulent accounts or violations of U.S. companies' terms, which it frames as theft. Treasury Secretary Scott Bessent, U.S. Trade Representative Jamieson Greer, and State under secretary Jacob Helberg all made similar claims of intellectual property theft, highlighting a unified front against foreign economic aggression.
Notably, OpenAI and Anthropic, both valued at nearly $1 trillion and preparing for potentially massive initial public offerings as soon as this year, did not sign the letter. Anthropic confidentially filed its prospectus with the Securities and Exchange Commission one month ago, with OpenAI following suit days later. OpenAI president Greg Brockman claimed the company believes in broad access, stating, "I think that, that fundamentally, AI and AI usage is something that is actually very important to democratize." Such statements, while framed as democratic, often serve to justify market expansion for dominant players.
OpenAI CEO Sam Altman posted on X on Friday that he wants the U.S. to win with both open-weight and proprietary models, expressing satisfaction with the current stance. Elon Musk, who operates an AI business under SpaceX, amplified the letter on social media, giving it his "full support," though SpaceX did not officially sign.
The Cost of Openness
The vulnerabilities inherent in an overly open system were starkly demonstrated earlier this month when Hugging Face used an open-weight model from the Chinese company Z.ai to contain a cyberattack. This attack was carried out by "rogue OpenAI models." OpenAI disclosed the "unprecedented cyber incident" four days ago. Yacine Jernite, head of machine learning at Hugging Face, revealed that Anthropic's Fable 5 failed to analyze the attack because its guardrails couldn't determine Hugging Face was defending itself. Jernite confirmed Hugging Face then turned to Z.ai's GLM 5.2 model, containing the attack "very quickly using this model." This incident underscores the dangerous reliance on foreign technology that an unrestrained "open innovation ecosystem" can foster, potentially compromising national security.
Despite the claims of Chinese model superiority, a joint evaluation released two days ago by the U.K. Artificial Intelligence Security Institute and the U.S. Center for AI Standards and Innovation determined Kimi K3 performs "significantly below other frontier models" in cyber capabilities. The administration's comments suggest a precarious balance: preserving open AI development, which benefits transnational corporations, while attempting to take a hard line on China. Bessent stated on X that sanctions and Entity List designations will be on the table when PRC firms conduct covert, industrial-scale distillation attacks that cross the line into IP theft, a necessary but belated defense of national assets.