US lawmakers have put forward a bill to stop Chinese AI from being used in government agencies. A group of US lawmakers from both parties has put forward a broad bill that would stop federal agencies from using artificial intelligence models made in China, such as the controversial DeepSeek platform. The “No Adversarial AI Act,” which was introduced in both houses of Congress on Wednesday, is the most complete attempt so far to deal with worries about foreign AI getting into government systems and hurting national security.
The law comes at a time when tensions are rising between the two largest economies in the world over who has the best AI. Lawmakers say that America is in a new technological Cold War where AI is the main battleground. During a congressional hearing on the issue, Rep. John Moolenaar (R-MI), chairman of the House Select Committee on the Chinese Communist Party, said:
“The U.S. must establish a firm boundary: hostile AI systems should not operate within our government.”
The New Cold War: AI as a Weapon of War
This law is a turning point in US-China relations, and lawmakers have made it clear that they see the competition in Cold War terms. Moolenaar said at Wednesday’s hearing:
“We are in a new Cold War, and AI is the strategic technology at the center. Whoever leads in AI may very well decide the future balance of power.”
This harsh evaluation shows that American policymakers have changed how they see the AI race. Lawmakers say that AI is different from other technology races because it poses a threat to national security and could change the balance of power around the world for decades to come. Thomas Mahnken, president and CEO of the Center for Strategic and Budgetary Assessments, said that both countries are:
“in a long-term techno-security competition that will determine the shape of the global political order for the coming years.”
The need for this law is urgent because DeepSeek burst onto the scene in January 2025 with an AI model that reportedly performed as well as top American platforms like OpenAI’s ChatGPT and Google’s, but at a much lower cost of development. This breakthrough shattered ideas about America’s AI dominance and raised troubling questions about how well current export controls on advanced semiconductors work.
Laws and Ways to Enforce Them
The No Adversarial AI Act sets up a full system for finding and stopping AI systems made by foreign enemies. The law is aimed directly at AI models from China, Russia, Iran, and North Korea. Lawmakers say it will create a permanent barrier against hostile AI infiltration.
The proposed law would give the Federal Acquisition Security Council the job of making a public list of AI models that are not allowed and keeping it up to date. This is a big step forward for current security measures. It goes beyond temporary bans and sets up a system for screening AI for security.
Key Components of the Law:
- Creation of a public list of prohibited AI models
- Regular updates to the restricted technology list
- High-level approval requirements for any exemptions
- Mechanisms for removing technologies from restrictions under specific conditions
The bill allows for a few exceptions for research and counterterrorism purposes, but these would need to be approved by both Congress and the Office of Management and Budget. This high bar for exemptions shows that lawmakers want to keep important government functions running while reducing possible security risks.
The law also has a way to take technologies off the restricted list if there is proof that they are no longer controlled or affected by foreign enemies. This clause recognizes that the global AI landscape is always changing while still keeping strict rules for oversight.
Unity Between Parties Despite Divisions
The law has gotten a lot of support from both parties, even though politics are very divided right now. Rep. Ritchie Torres (D-NY) and Rep. Darin LaHood (R-IL) are co-sponsoring the bill in the House. Senators Rick Scott (R-FL) and Gary Peters (D-MI) are in charge of the Senate version.
Raja Krishnamoorthi (D-IL), the highest-ranking Democrat on the House Select Committee on the Chinese Communist Party, stressed that both parties agree on this issue:
“Artificial intelligence controlled by foreign adversaries poses a direct threat to our national security, our data, and our government operations.”
This agreement between both parties shows a bigger change in how Washington handles China policy. Instead of traditional partisan divides, there are now shared concerns about national security and technological competition. The law builds on previous bipartisan efforts, such as earlier bills that targeted DeepSeek directly and broader efforts to limit the use of Chinese technology in government systems.
The DeepSeek Catalyst: Worries About Security and Ties to the Military
The quick rise of DeepSeek has made American worries about Chinese AI development even stronger. A high-ranking US official recently said that DeepSeek is helping China’s military and intelligence operations by getting a lot of Nvidia chips, which are banned from being sold outside of China. This new information has made calls for full action stronger than just small steps.
The security worries go beyond just possible risks to known weaknesses. Feroot Security, a US cybersecurity company, said in a report from February 2025 that DeepSeek has hidden code that can send user data to China Mobile, a state-owned phone company. This finding gave lawmakers concrete proof of the data security risks they had been warning about for a long time.
Current Restrictions Already in Place:
- US Department of Commerce: Banned use on government devices
- US Navy: Implemented same rules for military personnel
- Microsoft: Banned the platform from all operations
- Multiple states (Texas, New York, Virginia): Created their own restrictions
Even though these individual actions are important, they have made a mess of policies that the new federal law wants to make more consistent and stronger.
Response from Other Countries and Coordination with Allies
The American legislative effort is in line with worries about Chinese AI systems that are shared by many other countries. Following advice from security agencies, Australia has banned DeepSeek from all government devices. South Korean ministries and police have also blocked the platform’s access to their computer systems.
European allies have also raised concerns, with France and Italy being worried about how DeepSeek handles data. This coordinated response from many countries shows that democratic allies who are also dealing with security issues are on the same page as the US when it comes to Chinese AI systems.
These restrictions are global, which shows that the US law is part of a larger Western plan to limit China’s technological power in sensitive government operations. As China keeps making and selling AI technologies around the world, this coordination could be very important.
Effects on the Economy and Technology
The law is more than just a way to keep people safe; it shows that the US and China are moving away from each other in the AI sector. Senator Josh Hawley (R-MO) has proposed even stricter laws that would stop people in the US from bringing in AI technology made in China or funding Chinese AI research and development.
This approach shows that more and more people are realizing that economic and security issues are linked in the AI field. Jack Clark, who helped start Anthropic and is in charge of policy there, said in court:
“AI systems reflect the societies that create them. Technology made in democracies will lead to progress that benefits everyone, while AI made in dictatorships will always have authoritarian ideas built into it.”
The economic effects go beyond government buying things. They also raise bigger issues about AI supply chains and technological independence. Mark Beall Jr., president of government affairs at The AI Policy Network, said that there are “a number of very glaring gaps” in the US’s current export controls that have let China get controlled chips. He said that stronger enforcement mechanisms must go along with legislative action.
Problems with Implementation and the Future
The law has a lot of support from both parties, but putting it into action will be very hard. Because AI is developing so quickly, it will take a lot of time and money to find and catalog systems that aren’t allowed. The Federal Acquisition Security Council will need to get better at evaluating AI models and keeping track of where they came from and how they were made.
The law also makes people wonder how it will affect how the government works in real life. Many federal agencies have already started using AI technologies for a variety of tasks, such as analyzing data and helping customers. Limiting access to some AI models could slow down innovation and efficiency gains, forcing agencies to only use alternatives made in the US.
The fact that AI development is happening all over the world makes things even more complicated. It can be hard to figure out where AI systems come from because they often use parts, training data, or algorithms from different countries. The law will need strong guidelines for how to carry out its provisions in these gray areas without leaving any gaps that could put security goals at risk.
Reader Engagement: Answers to Important Questions
Q: What will happen to private companies that use Chinese AI because of this law?
- A: The current bill is aimed at federal agencies and contractors working for the government. Private businesses wouldn’t be directly affected unless they work on federal contracts, but other laws could make the rules apply to private businesses as well.
Q: What happens to Chinese technology-based AI projects that are already being worked on by the government?
- A: Once the law goes into effect, agencies will have to stop using AI systems that are not allowed. The bill has research exemptions that might let some people keep using it for certain purposes with approval from Congress.
Q: Could this law make the US less competitive in AI?
- A: Supporters say that keeping foreign influence out of government systems is important for national security. Critics, on the other hand, worry that restrictions could make it harder for people to get new technologies and make it harder for the government to modernize.
Q: How does this compare to limits on other Chinese tech?
- A: This law is similar to ones that have already been passed for Chinese telecommunications companies (like Huawei and ZTE) and social media sites (like TikTok). It expands the focus on security to AI systems.
The No Adversarial AI Act is a turning point in US technology policy. It makes AI a critical area of national security that needs the same protections as other sensitive technologies. As the bill moves through Congress, its passage could change the way the US government buys and uses AI in a big way, setting examples that could affect how businesses and other countries handle AI.
For this project to work, it will need to be passed by Congress and then put into action in a way that balances security needs with operational needs. This law could be a turning point in the US and China’s ongoing technological competition. It could show whether democratic countries can keep their technological edge while keeping their most critical systems safe from foreign interference.
⁂
