A bipartisan group of U.S. lawmakers has proposed the “No Adversarial AI Act,” aiming to restrict the federal government’s use of artificial intelligence developed in countries considered geopolitical rivals, including China, Russia, Iran, and North Korea.

The bill was introduced in the House by Republican John Moolenaar (Michigan) and Democrat Raja Krishnamoorthi (Illinois), both leaders of the House Select Committee on the Chinese Communist Party. A companion bill is also being led in the Senate by Senators Rick Scott (Florida) and Gary Peters (Michigan).

Main Provisions of the Bill:

Prohibits federal agencies from acquiring or deploying AI models originating from the targeted countries.
Mandates the Federal Acquisition Security Council to maintain and update a list of such AI systems.
Allows exemptions only with approval from Congress or the Office of Management and Budget.
Creates a process to remove AI models that are found not to be under foreign adversarial control.
Focus on DeepSeek and National Security Concerns

The proposal arises amid concerns over Chinese-based AI firm DeepSeek, which has developed a ChatGPT-like system at a lower cost. U.S. intelligence agencies have linked DeepSeek to support for China’s military and intelligence operations, as well as access to advanced Nvidia chips. Several agencies and companies have already restricted or banned DeepSeek products, and discussions of formal restrictions have been reported under the previous Trump administration.

Lawmakers’ Perspectives

Representative Moolenaar highlighted the importance of protecting federal networks from foreign surveillance and interference, emphasizing the need for a clear boundary against hostile AI systems. The bill enjoys support from both parties, including Democrat Ritchie Torres and Republican Darin LaHood, with aims to develop a comprehensive, long-term strategy for managing risks associated with foreign AI technologies in government systems.