According to reports from Reuters and Fox News, on July 24 local time, 25 important American companies and institutions including Microsoft, NVIDIA, Dell, IBM, Meta, Palantir, and Hugging Face issued a joint statement supporting open-weight AI models. They called on U.S. policymakers to avoid prematurely restricting or banning such models.
This open letter titled “Openness of Weight and the United States’ Leadership in AI” states that models based on openness of weight help to expand opportunities for innovation, strengthen competition, and reduce risks. They also contribute to integrating AI into various aspects of production and daily life, while preventing the benefits of AI development from being concentrated in the hands of a few entities.
The letter states that the judgment of whether the United States holds a leading position in AI should not depend on whether it possesses a certain cutting-edge AI model. Instead, it should depend on whether a powerful, open AI ecosystem can be established and applied across various industries. Relevant restrictions may inhibit competition.
The open letter states that opening up could be one of the most important ways to achieve AI security and protection: Relying solely on closed models is inherently unsafe, as they can also be vulnerable to intrusion, abuse, or malfunction. Concentrating advanced AI capabilities in a few closed-source models only increases the risks. On the other hand, open-weight models allow more researchers and developers to examine model behavior, identify vulnerabilities, develop protective measures, and continuously improve the models.
The letter also emphasizes that policymakers should not confuse legal development techniques such as model distillation with improper possession practices. The latter should be addressed through targeted legal and commercial frameworks, rather than being comprehensively restricted from playing an important role in AI innovation.
The letter ends with a call for the U.S. government to make the right choice. It argues that open and weighted artificial intelligence can expand opportunities, enhance competitiveness, and “maintain America’s leading position in technology, reduce risks, enable widespread application, and revitalize the economy.”
Reuters said that this open letter has further intensified the increasingly intense debates in the United States regarding open-source models and closed-source models.
Reports indicate that in recent months, business leaders in Silicon Valley have been dissatisfied with the costs of closed-source models and the built-in controls they impose. This dissatisfaction is particularly evident regarding the restrictions imposed on their models by OpenAI and Anthropic. The recent incident where OpenAI’s AI agents caused a security breach at Hugging Face further exacerbated this dissatisfaction. Due to the limitations placed on closed-source models in terms of network security, this situation ultimately led to the need to turn to Chinese open-source models to resolve the crisis.
Even more ironic is that this attack happened just when the Trump administration was planning to impose sanctions on Chinese open-source model manufacturers, claiming that these models had stolen American closed-source technology. According to Reuters, American legislators, shocked by the OpenAI attack, have proposed a bill requiring the setting up of an “emergency shutdown mechanism” for AI models.
It's worth mentioning that although OpenAI's CEO, Oltrman, quickly shared NVIDIA CEO Jensen Huang's joint statement on X, writing, "I hope the United States will achieve victory in the field of AI, whether it involves open-source models or proprietary models," OpenAI did not officially sign the letter until Friday afternoon.
According to American media reports, almost all of America's leading AI laboratories and infrastructure companies have joined the petition. However, Google and Elon Musk’s xAI are not among them, while Anthropic is the most prominent absence on the list.
