Meta, Microsoft, Nvidia, and IBM support open-weight AI

Two dozen companies and organisations have signed an open letter asking US policymakers to protect openweight AI models. The signatories include major technology firms such as Meta, Microsoft, Nvidia, IBM, Dell Technologies, and CrowdStrike, along with groups like Hugging Face, Mozilla, the Linux Foundation, and several investors. The letter argues that openweight models, whose trained parameters are available for download and local use, can reduce costs, broaden competition, and help organisations avoid dependence on a single vendor. It also says that businesses, researchers, and public institutions could use these models more flexibly than closed systems offered only through APIs. The signatories acknowledge security risks, including the possibility that released weights could be modified or redistributed without safety controls. Still, they argue that open models can improve security research by allowing more teams to inspect, test, and redteam systems, and they say concerns about misuse should be addressed through targeted rules rather than broad restrictions. The letter also defends distillation as a normal machine learning technique, while drawing a distinction between legitimate model development and the unauthorised extraction of value from closed systems. It arrives as a policy signal to Washington, where debates over AI regulation, compute access, and model release rules are still unresolved.