A bipartisan group of US lawmakers introduced a bill on June 25, 2025, to prohibit federal agencies from using AI models developed in countries such as China, Russia, Iran, and North Korea.

The legislation, named the “No Adversarial AI Act,” was presented in both the House of Representatives and the Senate.

The bill follows national security concerns over AI systems such as DeepSeek, a Chinese company reportedly linked to military and intelligence activities.

US companies and government agencies have previously acted to ban DeepSeek’s AI models due to data security concerns.

The House bill, led by Representatives John Moolenaar and Raja Krishnamoorthi, proposes a restricted list of foreign AI models, requiring congressional or Office of Management and Budget (OMB) approval for any federal use.

It allows for removal from the restricted list if there’s proof the technology is free from foreign control or influence.

Representative Moolenaar said that the legislation aims to protect US government networks from hostile AI systems.

Senators Rick Scott and Gary Peters introduced the Senate version, with support from Representatives Ritchie Torres and Darin LaHood in the House.

.source-ref{font-size:0.85em;color:#666;display:block;margin-top:1em;}a.ask-tia-citation-link:hover{color:#11628d !important;background:#e9f6f5 !important;border-color:#11628d !important;text-decoration:none !important;}@media only screen and (min-width:768px){a.ask-tia-citation-link{font-size:11px !important;}}

🔗 Source: Reuters

🧠 Food for thought

1️⃣ Tech decoupling accelerates with each regulatory action

The new bill barring U.S. agencies from using Chinese AI models represents a significant escalation in the ongoing technological separation between the U.S. and China.

This legislation comes amid a broader shift in U.S. tech policy from risk management to strategic pressure, as evidenced by expanding export restrictions on advanced technologies like NVIDIA’s chips 1.

The focus on DeepSeek illustrates how the U.S. is now targeting not just hardware but also AI software and services, creating a more comprehensive technological firewall.

This pattern of increasing restrictions has forced Chinese companies to adapt their operations, with many facing significant challenges due to tightening access to Western technology 1.

The bill’s permanent framework approach signals that tech decoupling is becoming institutionalized rather than remaining a temporary response to specific concerns.

2️⃣ China’s AI capabilities advance despite mounting restrictions

The legislative focus on DeepSeek comes as Chinese AI models are rapidly closing performance gaps with U.S. counterparts, even as access to advanced chips remains restricted.

Despite export controls, Chinese AI models like DeepSeek’s are now trailing U.S. models by only three to six months in capabilities 2, creating urgency for regulatory responses.

This progress reflects years of strategic planning and investment in AI infrastructure in China, with their approach characterized by strong institutional alignment between government, academia, and industry 3.

Export restrictions have spurred innovation and efficiency improvements in China’s AI sector 3, demonstrating the complex and sometimes counterproductive nature of tech containment policies.

The U.S. maintains advantages in total compute capacity and broader AI ecosystem development 4, but the narrowing performance gap highlights why policymakers are increasingly concerned about Chinese AI models in sensitive government applications.

……