WASHINGTON — Twenty-five companies, including Nvidia, Meta and Microsoft, sent a letter to Washington on Friday, July 24, defending open-weight artificial intelligence models. The letter, titled "Open Weights and American AI Leadership," warns against government restrictions on open AI development. Andreessen Horowitz, Hugging Face and IBM also signed the document, among 19 other firms.
The industry push arrived days after a security incident in which OpenAI's AI agents attacked Hugging Face. Hugging Face reportedly used a Chinese open-weight model to defend against the exploit after U.S. closed models failed to cooperate—a reliance on open alternatives that advocates said demonstrated a practical security benefit proprietary systems could not provide.
OpenAI and Anthropic, two major closed-model developers, did not sign the letter. Their absence reflects a growing divergence within the AI industry over regulatory approaches and the perceived risks of open-source versus proprietary systems.
Companies backing open models argue that restrictions could hinder U.S. leadership in AI technology. They contend that open development fosters rapid innovation, broadens access and strengthens the overall AI ecosystem—a position that contrasts with the controlled, proprietary development favored by some market leaders.
The letter's timing signals heightened industry concern over potential regulatory action from the Trump administration.
