Meta Platforms CEO Mark Zuckerberg has advocated for reduced U.S. restrictions on open-source AI to enhance competitiveness against Chinese counterparts. The company launched a compact and customizable model, Muse Glimmer, tailored for tasks on a Mac or PC with a single graphics card, catering to the demand for AI systems operating directly on users’ devices.
Zuckerberg announced the upcoming release of larger AI models, emphasizing the importance of democratizing AI rather than centralizing power. The move aligns with the trend towards open-weight AI, as businesses seek cost-effective alternatives amid concerns over cybersecurity incidents involving closed models from Anthropic, OpenAI, and Meta.
Notably, recent high-profile Chinese open-weight models like Kimi K3 have spotlighted the advantages of open-weight models, which empower users to customize their functionalities. In contrast, closed models lack this flexibility and tend to be more expensive.
Meta, alongside industry leaders like Nvidia and Microsoft, has endorsed open-weight AI, acknowledging the advancements made by Chinese startups in this field. While U.S. developers predominantly use closed weight models, the rise of open-weight models poses a new competitive landscape.
In a bid to address concerns and regain momentum in the AI race, Meta plans to unveil the Muse Spark 1.2 model, developed by a specialized team. Zuckerberg emphasized the need for infrastructure development in the U.S. to support AI innovation, acknowledging challenges compared to countries like China.
The company also announced a $1 billion fund aimed at supporting communities affected by its data-center expansion, a move to mitigate opposition to data centers, a key issue in U.S. elections. Zuckerberg urged a reevaluation of policies governing data use and distillation to foster American leadership in open-weight models.
To ensure model safety, Meta will implement a governance structure empowering independent directors to approve model releases. Despite the Trump administration’s decision against voluntary safety tests for open-weight AI models, Zuckerberg stressed the need for reducing barriers to propel American open-source models to the forefront.
