In testimony prepared for Congress, AI researcher Nathan Lambert argues that open-weight models have fundamentally altered the competitive landscape of artificial intelligence. The balance of power, he says, is no longer held exclusively by a handful of large labs. Instead, openly released models are enabling a wider set of organizations and individuals to build on and adapt advanced AI systems.
Lambert's analysis suggests that this shift carries both opportunity and risk. On one side, open models can spread capability and reduce dependence on dominant companies. On the other, they raise questions about oversight, safety, and who is accountable when powerful technology is widely available.
The testimony does not present a single policy answer, but it frames openness as a central variable in the future of AI governance. How regulators weigh the benefits of access against the risks of misuse, Lambert implies, will help determine whether the next phase of AI development is more concentrated or more distributed.