US lawmakers push new bill to ban DeepSeek and other Chinese AI models across gov’t agencies

US lawmakers on Wednesday have introduced a bill in both the House and Senate that would prohibit federal agencies from using artificial intelligence models developed in China, such as DeepSeek, as well as from Russia and other nations hostile to the US.

The bipartisan bill, dubbed the "No Adversarial AI Act," would create a permanent framework for barring all US federal agencies from using any Chinese, Russian, Iranian, or North Korean AI models. It would further ban the use of foreign adversarial models on government devices.

The proposed legislation was introduced into the US House of Representatives on Wednesday by Michigan Rep. John Moolenaar, a Republican and Chair of the Select Committee on the Chinese Communist Party (CCP), and Democrat Rep. Raja Krishnamoorthi of Illinois, a ranking member on the committee.

“We are in a new Cold War—and AI is the strategic technology at the center. The CCP doesn’t innovate—it steals, scales, and subverts," Moolenaar said in reference to the bill.

“From IP theft and chip smuggling to embedding AI in surveillance and military platforms, the Chinese Communist Party is racing to weaponize this technology. We must draw a clear line: US government systems cannot be powered by tools built to serve authoritarian interests,” he said.

Today, @CongressmanRaja and @RepMoolenaar led a bipartisan and bicameral bill, called the 𝐍𝐨 𝐀𝐝𝐯𝐞𝐫𝐬𝐚𝐫𝐢𝐚𝐥 𝐀𝐈 𝐀𝐜𝐭, which will protect federal agencies from the risks posed by AI technologies controlled by foreign adversaries.

Full Release: https://t.co/cHAVlZTH55

undefined Select Committee on the CCP Democrats (@CmteOnCCPDems) June 25, 2025

Chinese AI models rival US

"DeepSeek, as an example, has documented ties to the Chinese Communist Party and its intelligence apparatus, and its privacy policy explicitly states that US user data is stored in China," the lawmakers said in Wednesday's announcement.

Releasing its first large language model in November 2023, the Chinese-based AI company made waves in January, claiming its newest DeepSeek-R1 reasoning model was light years ahead in capabilities, and able to operate at a much lower cost compared to the most advanced US-made models, such as OpenAI's ChatGPT and Meta's Llama model.

This spring, US researchers discovered that the DeepSeek’s reasoning model could be use for nefarious purposes such as to generate ransomware and other malicious code.

Another powerful Chinese-developed model is Manus, an AI assistant released this March that is considered more advanced than the typical chatbots ChatGPT, Claude, and Grok. It was developed by the AI start-up company, Butterfly Effect, based in Beijing. And, the Chinese-owned search engine leader, Baidu, released its first AI reasoning model, the ERNIE X1, also in March.

What's in the bill?

The six-page bill would require the US Federal Acquisition Security Council to create and maintain a publicly available list of AI models developed in adversarial nations, which must be updated every 180 days.

Additionally, US federal agencies would not be able to purchase or use those AI technologies without an exemption from the US Congress or the Office of Management and Budget. Exemptions would include "a narrow exception for research, testing, or national security purposes" or to prove that the AI model is not controlled or influenced by a foreign adversary of the US.

“Artificial intelligence controlled by foreign adversaries poses a direct threat to our national security, our data, and our government operations,” said Krishnamoorthi. “This bipartisan legislation will create a clear firewall between foreign adversary AI and the US government, protecting our institutions and the American people,” he said.

The legislation also covers AI models that may emerge in the future, which could pose a similar or even greater threat to federal systems and data, the lawmakers said.

The move comes after a senior US official concluded that DeepSeek is aiding China's military and intelligence operations and has had access to "large volumes" of Nvidia's chips, according to a Reuters report.

Since then, some US companies and government agencies have banned the use of DeepSeek over data security concerns, while the Trump administration has also contemplated banning its use on US government devices.

Co-sponsors of the legislation include House members Ritchie Torres (D-NY) and Darin LaHood (R-IL) and Senators Rick Scott (R-FL) and Gary Peters (D-MI) in the Senate.