From 2022, America High-flying companies like OpenAI, Google DeepMind, Anthropic, and XAI have made solid gains in artificial intelligence thanks to advanced models. However, a growing number of experts are concerned that the US is beginning to lag behind when it comes to creating open-source AI models that can be downloaded, customized, and run locally.
Open models from Chinese companies like KM, Z.AI, Alibaba, and DeepSeek are now rapidly gaining popularity among researchers and engineers around the world, leaving the US behind in an increasingly important area of AI innovation. “The United States needs open models to solidify its lead at every level of the AI stack,” Nathan Lambert, founder of the ATOM (American Truly Open Models) project, told WIRED.
The most advanced models of American companies can be accessed only through a chatbot interface or by sending queries to the companies’ servers through an application programming interface or API. OpenAI and Google have released open-weight models, but they are much less capable than Chinese offerings, which are better suited to modification and provide more developer support. Chinese model makers also benefit from open-sourcing their models, as the best ideas and changes from outside researchers can be incorporated into future releases.
Lambert, who is also a researcher at the Allen Institute for AI (AI2), a nonprofit in Seattle, Washington, founded the ATOM project to highlight the risks associated with America lagging behind in open source. He says the country needs cutting-edge open models, because relying on foreign models could prove problematic if those models were suddenly shut down or made closed-source.
Lambert says open models also foster innovation and experimentation among startups and researchers. Additionally, companies with sensitive information need open models that they can run on their own hardware. “Open models are a fundamental part of AI research, dissemination, and innovation, and the US should take an active role rather than follow other contributors,” says Lambert.
The ATOM project, launched on July 4, presents a compelling argument for greater openness and shows how Chinese open-weighing models have overtaken the US in recent years.
Ironically, the open source AI movement was started by American social media giant Meta when it released Llama, an open-weighted frontier model, in July 2023. At the time, Meta saw Llama as a way to join the AI race. Very soon its new model became popular among researchers and entrepreneurs.
Since then, Meta and other US AI companies have ideally focused on the idea of developing human or superhuman-level AI before their competitors, resulting in less openness. In recent months, Zuckerberg has revamped Meta’s AI efforts with expensive hires and a new “superintelligence” lab. Zuckerberg has also indicated that Meta may no longer open-source its best models.
In contrast, China’s tech industry has moved toward greater openness this year. In January 2025, DeepSeek, a little-known startup, released an open model called DeepSeek-R1, which shocked the world due to its advanced capabilities and the fact that it was trained at a fraction of the cost of leading US models. Since then, several Chinese companies have introduced powerful open-weight models with additional innovations.