In a rapidly evolving landscape of artificial intelligence (AI), a coalition of 25 leading tech companies, including Nvidia, Microsoft, Meta, and Palantir, has issued a plea to policymakers to avoid imposing premature restrictions on open-weight AI models. These models, which can be downloaded, modified, and operated independently by users, have recently become a hot topic of debate within the technology sector.
The discussion was intensified as Chinese open-weight models emerged as strong contenders against established offerings from giants like OpenAI and Anthropic, who primarily focus on proprietary, closed models. This competitive shift has prompted officials and executives to consider whether limiting access to these Chinese models in the U.S. might be prudent.
Earlier this month, Moonshot AI, a Chinese startup, caught attention by releasing the Kimi K3 model, which surpasses some American models’ benchmarks. Following these developments, U.S. Treasury Secretary Scott Bessent stated that the Trump administration plans to investigate potential intellectual property theft by Chinese companies and is considering imposing sanctions.
Despite these tensions, the letter from U.S. tech companies emphasized the importance of open-weight models in maintaining healthy competition and distributing AI’s benefits widely. The companies warned that exclusively relying on closed models poses risks of breaches, misuse, and undetectable failures. Moreover, confining advanced AI capabilities to a few closed models could amplify these risks.
Notably absent from the signatories of this letter are OpenAI and Anthropic. Both companies are on the brink of significant initial public offerings (IPOs) and are each valued near a trillion dollars. OpenAI president Greg Brockman expressed support for broad AI access, underscoring its democratizing potential. He stated that expanding access to various models is fundamentally beneficial.
Elon Musk, who operates an AI division within SpaceX, echoed this sentiment by endorsing the letter on social media. SpaceX, however, did not officially sign the letter.
Concerns about security have been underscored by recent events. The AI company Hugging Face, following a cyberattack involving rogue OpenAI models, utilized a Chinese AI model to contain the threat effectively. This incident illustrated the varied and collaborative responses required to address the dynamic challenges within AI innovation.
White House advisor Michael Kratsios recently accused Moonshot AI of using a method known as “distillation” to develop its Kimi K3 model by leveraging Anthropic’s technology. In a social media post, Kratsios noted that while legitimate distillation is crucial for open innovation, large-scale industrial distillation intended for intellectual property theft is unacceptable.
The letter from the U.S. tech firms emphasized addressing unlawful distillation through targeted legal frameworks rather than broad restrictions, highlighting that the nation’s AI leadership should not hinge solely on individual advanced models but rather on a robust, open ecosystem that fosters innovation and growth across all sectors.
Read more about the ongoing discussions and the original article here: [https://www.cnbc.com/2026/07/24/nvidia-microsoft-meta-open-weight-ai-models.html](https://www.cnbc.com/2026/07/24/nvidia-microsoft-meta-open-weight-ai-models.html)
