AI stalwarts are backing open-weight AI models as the key to fair competition and lowering AI costs in an argument to lawmakers.

Top tech companies including Nvidia, Microsoft, Meta, and IBM wrote to lawmakers on Friday, urging them to halt “premature restrictions on open models that stifle competition or drive innovation overseas.”

The letter, which was posted on X, heightens the debate surrounding open source AI models and closed-source models.

Open-source models may be more difficult to regulate, but Palantir and Microsoft leaders both argue that they can help stymie out-of-control AI costs as firms can run the models in their own data centres.

Open-weight models can be downloaded and run, given access to the right compute power. The models’s training code and data are not, however, disclosed. China has largely taken the lead in developing these open-weight models, with its recent Kimi 3 model set to rival some of the top closed-source models developed in the US.

Open-weight models can be used by public organisations, businesses, researchers and universities without having to pay the price of a closed frontier model. These organisations can build on these models without building one from scratch.

The tech leaders also argued that open-weight models are integral to fair competition in AI “and competition is what keeps the gains of AI broadly shared rather than concentrated in a few hands.”

“Relying solely on closed models is not inherently safe: they can be breached, misused, or fail in ways that outsiders cannot detect,” the letter warns.

“Open ​weight models, on the ‌other hand, allow a broad community of researchers and developers to examine their ​behavior, identify vulnerabilities, develop safeguards, and improve them over time.”

This warning is perhaps best exemplified by the recent hacking of Hugging Face by OpenAI models.

Hugging Face said that it had to turn to an open-source Chinese AI model to help it identify and defend against the hack by OpenAI models due to the closed-source models’ restrictions on cybersecurity use.

The notion of a “rogue AI agent” has prompted US lawmakers to debate the creation of a “kill switch” for AI models, while the US president posits sanctions on Chinese open-source models as allegations that they copied closed-source US AI technology mount.

The letter addresses these concerns, however, saying that rather than “sweeping restrictions,”, lawmakers should target these harms through “legal and commercial frameworks.”

Recommended reading

• Anthropic Investigates Claimed Unauthorised Access to Mythos
• OpenAI Model Goes Rogue In Testing, Hacks Hugging Face
• 9 in 10 Cyber Leader Think Mythos Will Increase Cyber Risk
• Anthropic to Release Claude Mythos to the Public
• Is Mythos as Dangerous as Anthropic Claims?

Twenty-five firms signed the letter, with omissions including OpenAI, Anthropic, and Google.

However, Sam Altman, CEO OpenAI, supported the letter on social media, saying: “I want the US to win in AI both in open source and proprietary models, and I am glad to see this.”

Google CEO Sundar Pichai agreed, posting: “Very happy to support this on behalf of Google. We have long benefited from open source, are big contributors to open source and in fact have consistently made open weights models with Gemma available from @GoogleDeepMind and [DeepMind CEO Demis Hassabis]. Onwards!”