Meta Platforms CEO Mark Zuckerberg has urged U.S. policymakers to lower regulatory barriers for open-source AI, warning that excessive friction risks handing the competitive lead to Chinese rivals. The call to action coincided with Meta’s release of Muse Glimmer—a compact open-weight AI model optimized to execute agentic tasks directly on personal computers with a single GPU. Zuckerberg emphasized that larger systems will follow soon, framing open AI deployment as essential to avoiding an extreme concentration of power within a small group of tech entities.
Zuckerberg outlined his vision in a 14-page essay titled “The Future is for Everyone,” published alongside announcements of a $1 billion community fund designed to address local opposition to data center construction—an issue he cited as a major bottleneck for U.S. infrastructure expansion. Meta’s capital expenditure on AI infrastructure is projected to reach up to $145 billion this year. To address safety and governance concerns surrounding open releases, Meta plans to grant its independent board directors oversight authority to approve safety criteria prior to model deployment.
The advocacy for open-weight systems comes as Chinese developers, including Moonshot, Alibaba, and DeepSeek, rapidly advance open models that rival leading closed U.S. systems from OpenAI, Anthropic, and Google. Open-weight models are increasingly favored by enterprise clients seeking cost efficiency, customization, and operational autonomy, particularly following recent cybersecurity incidents. To maintain American leadership, Zuckerberg called for policy reforms around training data access and model distillation, noting that foreign competitors currently operate with fewer data restrictions. Meta also confirmed plans to release the weights for its flagship Muse Spark 1.2 model, developed by its superintelligence unit formed last year.
Market reaction to the announcement was positive, with Meta’s shares rising nearly 3% in premarket trading on Monday following a roughly 10% year-to-date decline. The policy debate surrounding open AI models remains active in Washington, where the Trump administration recently informed developers that open-weight systems will not be subject to voluntary safety evaluations.
Click here for more on Technology
