White House AI crackdown opens door for Chinese model makers to close gap

The recent restrictions imposed by the Trump administration on Anthropic's advanced artificial intelligence models may inadvertently benefit China, a significant competitor in the AI landscape. After a two-week hiatus due to export control measures, the White House granted Anthropic permission to introduce its robust Mythos 5 model to select companies and federal agencies. However, the Fable 5 model continues to remain unavailable. In a parallel move, OpenAI announced it would also curtail the deployment of its GPT 5.6 models at the request of the government. This tug-of-war among leading U.S. AI developers, including tech giants like Google, underscores the competitive race to innovate in AI technology. Industry leaders argue that excessive regulation could hinder domestic advancements, allowing China to close the gap rapidly. As Anthropic navigates national security concerns, Chinese developers are making strides with models that rival those of leading U.S. labs. The recent release of Zhipu's GLM 5.2 model has garnered attention for its performance, reportedly matching top U.S. capabilities in certain cyber benchmarks. Venture capitalist Marc Andreessen remarked on social media that GLM-5.2 represents a significant advancement for Chinese AI, often surpassing U.S. public models without compromise. Experts are sounding alarms about these developments. Sam Bresnick, a research fellow at Georgetown's Center for Security and Emerging Technology, labeled the situation a crucial wake-up call. Jefferies strategist Christopher Wood indicated that GLM 5.2 could compete effectively in the corporate market, offering comparable performance to Anthropic at only a quarter of the cost. Even David Sacks, a former advisor on crypto and AI, highlighted the alarming pace of China's advancements, suggesting that the U.S. must adhere to a pro-innovation strategy to secure its lead in the global AI race. The shift in corporate America is palpable. Companies are transitioning from a phase of unrestricted AI spending to a more cautious approach focused on efficiency. This change plays directly into China's hands, as demonstrated by AI startup Lindy’s recent decision to switch entirely from Anthropic’s models to DeepSeek, a Chinese firm known for its more affordable alternatives. The ease of access to open-weight models from Chinese developers has made it simple for U.S. companies to deploy these technologies on their servers, bypassing the need for third-party cloud services. Experts like Travis Lanham from Armadin noted that these models are showing promise for cybersecurity applications, which raises concerns about potential vulnerabilities. As U.S. policymakers grapple with how to manage this evolving landscape, the focus remains on safeguarding national security while fostering innovation. For years, the U.S. government has enforced strict export controls to prevent cutting-edge AI technology from reaching China. However, recent developments, including the clearance of Nvidia's H200 chip for export to China, complicate this narrative. With figures like Elon Musk predicting rapid advancements in Chinese models, the urgency for U.S. industry leaders to adapt is clear. Major corporations, including Shopify and Coinbase, are already leveraging Chinese models to reduce costs, emphasizing the growing influence of these technologies. Cybersecurity remains a top concern as the capabilities of open-weight models evolve. Experts warn that if the U.S. does not seize the opportunity to prepare, the imminent arrival of equally advanced Chinese models could leave the nation vulnerable to potential threats.