Banks Outlines Priorities for Pentagon’s U.S.—China AI Strategy as Tech Rivalry Intensifies

WASHINGTON, D.C. - Yesterday, Senator Jim Banks (R-Ind.) sent a letter to Secretary of War Pete Hegseth emphasizing the importance of the Artificial Intelligence Futures Steering Committee in the context of U.S.-China AI competition.

Read more about the letter here.

Read the full letter here or see below:

Dear Secretary Hegseth:

I am writing regarding the Artificial Intelligence Futures Steering Committee and its significance for
U.S.-China AI competition.

The FY2026 National Defense Authorization Act (NDAA) directs the Department of War to form a
senior-level panel to examine critical questions about advanced artificial intelligence (AI). The NDAA tasks the AI Futures Steering Committee to study advanced AI systems that could be developed in the upcoming years, form proactive policies for evaluating and adopting such systems, and analyze the trajectories of adversaries toward artificial general intelligence.

AI is likely to be the defining technology of the 21st century. Many experts predict that AI may soon
match or surpass human performance across various tasks, including AI research and development
itself. We do not know if this progress will occur rapidly or slowly, and it is wise for the Department to prepare for a variety of possibilities.

The People's Republic of China (PRC) is clearly aware of AI's potential. Last year, DeepSeek caught
much of our national security community by surprise. As AI competition intensifies, we need to take
appropriate actions to avoid strategic surprises. We must accurately assess the PRC's AI capabilities
relative to our own, understand the most prominent PRC AI developers, assess how key officials think about opportunities and risks from AI, and reliably detect PRC efforts to develop advanced AI.

The Steering Committee is well-suited to address several specific questions relating to PRC AI
development and U.S.-PRC AI competition. I suggest considering the following areas of focus:

• Assess PRC frontier AI progress. Who are the entities (companies, scientists, compute providers) within the PRC that are most critical for their advanced AI development? How effectively can we predict the capabilities of PRC frontier AI systems before they are released? How effectively can we estimate the gap between PRC and U.S. AI capabilities, and what information or capabilities would we need to have more precise estimates? What is the Department's current estimate of the U.S.-PRC frontier capability gap in key areas such as model performance and aggregate training compute?
• Analyze how the PRC views the future of advanced AI. Who are the scientists and policymakers shaping discussions around the future of AI in the PRC? To what extent are their beliefs similar to those of our scientists, industry leaders, and policymakers? For example, how does the PRC conceptualize the opportunities and risks from so-called artificial general intelligence or artificial superintelligence?
• Examine the potential for rapid capability amplification. Could AI systems become so capable at AI research and development tasks that we experience rapid capability amplification (also called "recursive improvement")? If so, could this allow one side to achieve an enduring AI lead over the other? Are PRC leaders aiming for AI-enabled rapid capability amplification, and how would we detect if they were close to achieving this?
• Review the security practices of leading PRC AI developers. How do the frontier PRC AI developers monitor or examine national security properties of AI systems? How do their security practices compare to the practices of U.S. AI developers? How do their security frameworks compare to those of U.S. developers (e.g., OpenAI's preparedness framework, xAI's risk management framework)?
• Understand methods for sabotaging or poisoning frontier AI development. If the PRC were to attempt to disable, sabotage, or manipulate the frontier AI development of the United States, what methods would they be most likely to use? What methods might be uncovered in the upcoming years, and what defenses exist to counter such methods?
• Investigate methods for detecting covert AI development or deployment. If the PRC secretly developed or deployed advanced AI systems in secret data centers, how well would we be able to catch them? What tools do we have to detect covert AI development or deployment? What indicators (e.g., power, cooling, chip flows, facility patterns) would we need to track, and what strategies might the PRC use to hide such indicators?

Several of these priority areas will involve coordination with the intelligence community. For these, the Steering Committee can play an important role by setting clear intelligence requirements, identifying the questions of greatest strategic importance, and shaping collection and analysis priorities accordingly. It will also be essential to ensure that resulting assessments directly inform the
Department's AI strategy, investments, policies, and warfighting tactics.

To support the successful implementation of the Artificial Intelligence Futures Steering Committee, I
request a staff-level briefing within 60 days to discuss the Department's plans for the Committee and
its role in U.S.-PRC AI competition. I look forward to following the work of the Steering Committee,
and I appreciate your work on this critical topic.

###