Meta Platforms CEO Mark Zuckerberg has advocated for reduced U.S. barriers on open-source AI to enhance competition with Chinese counterparts. This move comes as the social media giant unveiled a smaller, more customizable model called Muse Glimmer, tailored for tasks on individual computers using a single graphics card. Zuckerberg hinted at the imminent release of even larger models in a video post accompanying a 14-page essay promoting wider access to AI.

The spotlight on open-weight AI has intensified following concerns over rising costs and recent cybersecurity breaches involving models from Anthropic, OpenAI, and Meta. Open-weight models, in contrast to closed models, expose their core components, allowing users to tailor them for specific needs at a lower cost.

Several high-profile Chinese open-weight models, including Kimi K3, have gained traction, prompting American tech companies like Meta to rally behind open-weight AI. While U.S. developers like OpenAI and Google maintain closed weight models, Chinese startups have excelled in the open-weight model domain.

Meta’s shares, which have declined by around 10% this year, saw a nearly 3% uptick in premarket trading on Monday. The company plans to unveil Muse Spark 1.2, its advanced model developed by a specialized team. Zuckerberg also announced a $1 billion fund to alleviate concerns over data center expansion impacts.

In a bid to address infrastructure challenges and foster innovation in open-weight models within the U.S., Zuckerberg emphasized the need for policy reforms on data use and distillation. Meta aims to empower its independent directors to approve safety criteria for model releases.

The U.S. administration’s decision not to subject open-weight AI models to voluntary safety tests has sparked discussions on the need to streamline regulations to bolster American leadership in open-source AI development.