Securing AI Labs in the U.S. is a National Imperative

The United States is at a critical juncture with artificial intelligence as it scales to become capable across a wide range of applications. If current trends continue, advanced AI will transform both national security and economic competitiveness. And as U.S. tech companies and federal laboratories alike continue to push the frontiers of AI capabilities, they become prime targets of espionage, particularly from adversarial nations like China. Congressional leadership can increase protections against foreign espionage as they finalize the National Defense Authorization Act for 2025 to counter this strategic threat. Our adversaries are capitalizing on AI advancements.  The time is now for us to act to close the gaps in our security.

Currently, the Department of Energy oversees 17 federal laboratories, like Los Alamos, Lawrence Livermore, and Oak Ridge. These are the jewels in America’s research crown, and they often work on global research initiatives, especially in nuclear physics, AI, and materials science. Many foreign scientists, including those from countries with security concerns, have been granted access to these labs under collaboration agreements. According to the Government Accountability Office findings, and noted in the current Senate Intelligence Authorization Act, approximately 40,000 citizens of foreign countries, including more than 8,000 citizens from China and Russia, were granted access to the premises, information, or technology of National Laboratories in fiscal year 2023. This should be a cause for concern with global research initiatives that are centered around nuclear physics, AI, and materials science. 

 As conferees develop the final National Defense Authorization Act for FY25, it is critical to implement heightened protections for national labs through coordination with counterintelligence officials. Intelligence experts can determine the security risks and background check requirements for lab employees and visitors- for both physical and virtual access. We should not miss the opportunity to address vulnerabilities in federal and private AI labs to foreign espionage as Congress also continues to debate the merits of closing the loopholes China exploits in export control and outbound investment laws. When China gains access to advanced AI technologies through espionage, it not only enhances its military and intelligence capabilities but also enables more sophisticated cyberattacks on critical U.S. infrastructure, such as energy grids and communication networks. These acquisitions accelerate China’s ability to close the gap in military technology, eroding the conventional military advantages the United States currently holds. 

The National Defense Authorization Act should direct increased collaboration between AI labs and U.S. counterintelligence officials and close loopholes in existing export control policy to protect cutting-edge research. U.S. leadership in AI hinges on securing the intellectual and technological assets that drive it. The stakes are immense: AI is not merely a tool of economic growth; it is a linchpin of national security and geopolitical power. Safeguarding our federal and private AI industry from adversarial threats must be treated with the urgency it deserves, as it directly impacts our economic and military competitiveness. The global AI race is not just about who builds the most advanced AI capabilities—it's about who can protect and harness those advances to secure a better and safer future.

Michael Groen (LtGen, USMC, ret.) Is a former director of DoD's Joint Artificial Intelligence Center and a former senior Intelligence and Counterintelligence Officer.