A coalition of leading global technology companies, including Nvidia, Microsoft, Meta and IBM, has called on United States lawmakers to avoid imposing broad restrictions on open-weight artificial intelligence (AI) models, warning that premature regulation could weaken innovation, reduce competition and push AI development overseas.
The appeal comes amid growing debate over AI safety, cybersecurity and national security as governments around the world consider how best to regulate increasingly powerful AI systems.
In an open letter posted on X by Jensen Huang, the founder and chief executive officer of Nvidia, more than 20 technology companies and organisations argued that policymakers should pursue targeted regulations rather than broad measures that could hinder the development of open AI models.
The signatories cautioned against “premature restrictions on open models that stifle competition or drive innovation overseas,” insisting that the United States’ leadership in AI would depend on building an open and competitive ecosystem rather than relying on a handful of proprietary frontier models.
The letter was signed by organisations including Meta, Microsoft, IBM, Dell Technologies, Mozilla, Hugging Face, Mistral, Palantir, Replit, ServiceNow, CrowdStrike, Y Combinator, Andreessen Horowitz and The Linux Foundation, among others.
The coalition argued that open-weight AI models, whose trained parameters are publicly available for developers to download, inspect, modify and deploy on their own infrastructure, have become an essential foundation for expanding AI adoption across industries.
According to the group, such models enable startups, universities, businesses and public institutions to build advanced AI applications without the enormous costs associated with training frontier models from scratch.
They noted that wider access to open-weight models would make AI more economically sustainable by allowing organisations to deploy specialised models for everyday tasks while reserving the most advanced systems for complex problems.
The companies further argued that open AI promotes competition throughout the technology value chain, spanning model development, cloud infrastructure, chips, applications and services, ultimately driving innovation, lowering costs and broadening access to AI capabilities.
Beyond competition, the coalition said open-weight models give organisations greater control over their AI deployments by reducing dependence on a single technology provider.
It explained that businesses increasingly want the flexibility to manage their own data, customise AI models for specific needs and deploy them wherever operational requirements demand, while retaining the knowledge and capabilities they develop internally.
While acknowledging that open-weight models carry genuine risks because once released they can be modified beyond the control of their original developers, the companies argued that restricting access would not improve security.
Instead, they maintained that cybersecurity professionals require access to advanced AI models to detect threats, simulate attacks and strengthen defences against increasingly sophisticated cybercriminals.
The letter also challenged the notion that closed AI models are inherently safer, arguing that concentrating advanced AI capabilities among a small number of providers creates single points of failure and limits independent scrutiny.
The coalition contended that allowing researchers and developers to examine open models encourages vulnerability discovery, rigorous evaluation, red-teaming and continuous improvements, drawing parallels with the security benefits long associated with open-source software.
The signatories urged policymakers to support a strong AI ecosystem by expanding access to computing infrastructure for startups and researchers, investing in shared AI resources such as datasets, tools and evaluation frameworks, and encouraging the development of robust application layers that can accelerate AI adoption across the economy.
They also called on lawmakers not to confuse legitimate AI development techniques with intellectual property theft.
The letter specifically defended model distillation, a widely used process in which the outputs of one model are used to help train or improve another. According to the coalition, distillation has long been recognised as a legitimate method for model evaluation and improvement and should not be restricted because of concerns over unlawful extraction of value from proprietary models.
Instead, they recommended that disputes involving intellectual property or commercial misuse be addressed through targeted legal and commercial frameworks rather than sweeping regulations that could undermine AI innovation.
Concluding the letter, the coalition said AI presents an opportunity to expand economic prosperity if governments adopt policies that encourage openness and competition.
“With the right choices, open weight AI can expand opportunity, strengthen competition, extend American technological leadership, mitigate risk, and ensure that the benefits of this extraordinary technology are shared broadly across our economy,” the companies said.
The intervention comes as the debate over open-weight versus closed AI models intensifies globally. While proponents argue that open models accelerate innovation, transparency and accessibility, critics contend that their widespread availability could make powerful AI systems more difficult to regulate and increase the risk of misuse.
