On June 25, 2025, members of the US Congress from both political parties jointly introduced a significant bill that could reshape the future of AI competition between the United States and China. This legislation, named the “No Adversarial AI Act,” aims to prohibit US federal government agencies from using artificial intelligence (AI) models developed by adversarial countries, including China, Russia, Iran, and North Korea.
Why Ban Foreign AI?
The primary reason behind this move is growing concern over cybersecurity, particularly following reports that DeepSeek, a Chinese AI startup, has links to the Chinese military and intelligence agencies. Additionally, DeepSeek has reportedly gained access to significant amounts of critical hardware like chips from Nvidia.
DeepSeek gained widespread recognition in January 2024 after claiming that its AI model, DeepSeek-R1, outperformed models from US companies like ChatGPT and Llama while being far more cost-effective. More worryingly, researchers discovered that this model could be used to generate malicious code, such as ransomware.
DeepSeek is not the only Chinese AI under scrutiny. Several other emerging AI models from China, such as Manus from Butterfly Effect and ERNIE X1 from Baidu, have also made significant strides in the global market since early 2025.
Key Provisions of the "No Adversarial AI Act"
This bill outlines clear regulations for controlling the use of AI:
- Ban on AI from Rival Nations: Prohibits federal government agencies from using AI developed in designated adversarial countries: China, Russia, Iran, and North Korea.
- Establish and Update Adversarial AI List: Mandates the Federal Acquisition Security Council to create and maintain a list of AI models developed in these countries, updating it every 180 days.
- Require Special Approval: The use of AI from countries on the list will require special approval, such as for research, testing, or national security purposes.
- Mechanism for Removal: If there is evidence that an AI is not controlled by a foreign government, a request can be made to remove it from the list.
The "New Cold War" Driven by AI
This movement reflects a growing understanding that AI is the "strategic technology at the center" of a "new Cold War" between the US and China. Representative John Moolenaar, Chairman of the House Select Committee on China, stated that the future global balance of power may very well be determined by who leads in AI.
Many experts agree that this competition is a "long-term techno-security competition" that will shape the global political order in the years to come. Jack Clark, co-founder of Anthropic, has opined that AI developed in democracies will lead to better technology for humanity, while AI created in authoritarian nations will be inextricably linked to authoritarianism.
While the 2025 AI Index Report by Stanford University indicates that the US leads in producing top AI models, it also points out that China is rapidly closing the performance gap and is a leader in AI research publications and patents.
Is DeepSeek Banned Now?
Although this bill has not yet been enacted into law, some federal agencies and several US states have already banned the use of DeepSeek on government devices. The push for the "No Adversarial AI Act" aims to create a "permanent firewall" to completely exclude AI from rival nations from government systems.
The No Adversarial AI Act is not just about banning technology from China; it reflects drawing a "clear line" between the world of AI built on democratic values and AI developed under authoritarian regimes. In a world where AI is becoming central to every industry, controlling the source of this technology also means controlling power in the future.