The U.S. Congress is advancing legislation aimed at securing America's global leadership in the open weight artificial intelligence (AI) sector. This move comes as Chinese companies rapidly expand their influence in the open weight AI market by publicly sharing model weights, prompting the U.S. to support the development and dissemination of American models.
U.S. House Considers 'Open Source AI Leadership Act'
According to the South China Morning Post, the U.S. House of Representatives reviewed the 'Open Source AI Leadership Act' on September 1. The bill proposes establishing a dedicated office within the U.S. Department of Commerce to promote the use of American open weight AI models and to identify barriers to their adoption. It also mandates the collection of data to compare the utilization of U.S. and Chinese AI models.
Once enacted, the bill requires the Department of Commerce to disclose risks associated with the use of Chinese AI models, including censorship, supply chain vulnerabilities, and potential misuse, within 18 months.
Representative Gabe Evans, a Republican who introduced the bill, stated, "Unfortunately, the leading companies in the open source and open weight model space are Alibaba, DeepSeek, and Moonshot," claiming these companies are "deeply intertwined with the Chinese Communist Party." He noted that American companies are increasingly falling behind their Chinese counterparts in the global open weight AI market.
The bill passed to the full committee review stage through a verbal vote on the same day. If approved, it will proceed to a vote in the House.
This legislative push comes as American companies like OpenAI and Anthropic have led the way in closed AI models, while Chinese firms have quickly gained traction in the open weight AI market.
According to the AI-focused newsletter Interconnects, downloads of Chinese open weight AI models exceeded 328 million last month, more than double that of American models. This surge is attributed to the release of powerful new products like Moonshot AI's 'Kimi K3' and DeepSeek's updated 'V4 Pro' model.
There has also been an increase in U.S. companies, including Purple City and Meta Platforms, testing Chinese open weight models directly.
Focus on Promoting American AI Rather Than Blocking Chinese Models
Notably, the bill includes provisions that prevent the Department of Commerce from banning or restricting the distribution of open AI models. Instead of directly blocking Chinese AI models, the focus is on supporting the adoption and proliferation of American models.
Major U.S. tech companies, including NVIDIA, Microsoft, and Meta, have previously called for securing U.S. leadership in the open weight AI sector. However, they opposed the Trump administration's efforts to regulate Chinese open weight AI models, emphasizing the importance of fostering the development and spread of American models to enhance U.S. influence in the global AI ecosystem.
Meanwhile, as the U.S.-China AI competition intensifies, there are calls for cooperation in AI safety. Concerns about the potential for AI to operate beyond human control have emerged as a key issue.
Dean Ball, OpenAI's head of strategic futures and a former White House official, stated on September 31 that he believes U.S.-China AI safety cooperation is desirable and that a window of opportunity for collaboration is opening. He noted that the risks associated with AI and the upcoming Trump-Xi Jinping summit could lead to increased opportunities for cooperation in the coming months.
There is a possibility that discussions on AI safety measures will take place during the anticipated U.S.-China summit next month. U.S. Treasury Secretary Scott Bessenet also indicated in a previous interview with Reuters that authorities from both countries would continue dialogues on various topics, including AI safety measures and regulations.
However, while both the U.S. and China emphasize the importance of AI openness and safety, they differ in their perspectives. Recently, Chinese state media has voiced opposition to the U.S. regulating Chinese models under the pretext of AI safety or strengthening American technological hegemony.
China's state-run CCTV affiliate Yuyuantan Tian stated on August 30, "AI safety standards should not be unilaterally set by the U.S. or specific companies," arguing that labeling Chinese AI models as a 'security threat' confuses safety issues with technological competition.
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.