Sens. Bill Cassidy, R-La., and Jacky Rosen, D-Nev., have introduced the Protection Against Foreign Adversarial Artificial Intelligence Act of 2025 to protect federal data from adversarial artificial intelligence technologies, such as DeepSeek.
Under the proposal, companies would be prohibited from using DeepSeek to fulfill contracts with federal agencies, according to a Tuesday press release from Cassidy’s office. DeepSeek poses a potential national security threat as it is required by Chinese law to share the data it collects with the government, the statement added.
Table of Contents
Protecting Sensitive Federal Data
“AI is a powerful tool which can be used to enhance things like medicine and education. But in the wrong hands, it can be weaponized. By feeding sensitive data into systems like DeepSeek, we give China another weapon,” Cassidy said.
“This bipartisan legislation would prevent federal contractors from using Deepseek, a CCP-linked AI platform, when carrying out government work,” Rosen noted. “The U.S. must take steps to ensure Americans’ data and our government systems are protected against cyberthreats from foreign adversaries,” she stressed.
AI Bill Provisions
If enacted into law, the act would direct the secretary of commerce to work with the secretary of defense and submit a congressional report on the national security and economic espionage threats posed by AI platforms from China, North Korea, Iran and Russia. The measure would also authorize the secretary of commerce to issue a waiver for national security-related or research purposes.
In February, Rosen co-sponsored a related Senate bill that would ban DeepSeek from all government devices and networks to address concerns that the platform collects data and shares it with the Chinese government.