Nvidia, Microsoft, Meta warn against 'premature restrictions' of open-weight models

A group of 25 technology companies recently issued an open letter to government officials, arguing against premature restrictions on open-weight artificial intelligence models. Major industry players including Nvidia, Microsoft, Meta, and Palantir joined the effort. These businesses contend that keeping these models accessible is necessary to maintain competition and ensure broad technological access.

The debate centers on the rapid advancement of Chinese open-weight models, such as the Kimi K3 model from the startup Moonshot AI. Some U.S. government officials have discussed potential sanctions or restrictions on these models due to concerns about intellectual property theft and security. Treasury Secretary Scott Bessent noted that the administration is investigating these issues.

In their letter, the companies warned that blocking access to open-weight models could push innovation to other countries. They argued that closed models are not inherently safer and that consolidating power behind a small number of proprietary systems introduces new risks. Instead of broad bans, the group suggests using specific legal frameworks to address concerns about unauthorized use of technology.

Notably, OpenAI and Anthropic did not sign the letter. Both firms maintain a focus on proprietary models as they prepare for high-profile public offerings later this year. Despite the lack of an official signature from OpenAI, president Greg Brockman and CEO Sam Altman have expressed support for broad, democratic access to artificial intelligence.

The industry conversation remains urgent following a recent cyber incident involving Hugging Face, where the company successfully used a Chinese open-weight model to defend its systems after proprietary models failed to identify the attack. As policymakers evaluate their next steps, the tech industry continues to push for an open ecosystem that fosters growth across every sector in the United States.