Foreign policy in the United States relies on a mix of policy analysis, staffing processes, and bureaucratic procedures that have not changed for decades. The National Security Council (NSC), chartered to guide strategy formation and implementation, has proved impervious to advances in data science and artificial intelligence (AI). Instead, it relies on old-fashioned Microsoft Word documents and PowerPoint attachments to shape critical national security policy debates. This contrasts with the private sector, where analysts integrate dynamic and interactive tools like Notion, Jupyter Notebooks, or R Markdown files and other formats that can take real-time data to clean, transform, visualize, or model outcomes and communicate their meaning for evidence-based decisionmaking. Worse still, the government’s approach to strategy has yet to incorporate advances in AI.
The Trump administration has an opportunity to revolutionize the NSC by integrating AI into policy analysis and using new analytical techniques to create interagency playbooks that cut through the Byzantine federal bureaucracy. Taking this leap should go beyond publishing guidelines on the use of AI in government to building a set of analytical processes that take advantage of emerging technology to sustain the United States’ strategic advantage.
The inner workings of the foreign policy machine start with discussions during the transition that form the basis for a directive published in the first 30 days of a presidency outlining the structure and processes of the NSC. For example, the Memorandum on Renewing the National Security Council System was published on February 4, 2021, 15 days into the Biden administration. Similarly, the first Trump administration published a similar document just 8 days into his administration, on January 28, 2017. Setting the structure and processes early, this system took shape in the late 1980s, under President George H.W. Bush’s National Security Directive 1, published January 30, 1989.
These documents turn the National Security Council Act of 1947 into a bureaucratic reality that sees a symphony of policy coordination and interagency bargaining. Often the process changes are minor, with assistant security–level officials tasked to provide policy analysis for senior leader consideration based on a set of regional and topical issues directed by the NSC Deputies Committee. The structure changes are more significant, with core issues (e.g., competition with China, ending the war in Ukraine and reconstruction, and economic security) creating, in effect, a foreign policy focus and battle rhythm a president’s foreign policy agenda.
Regardless of the core issues around which the Trump team builds the NSC structure, the subsequent “National Security Memorandum 2” should take a bold leap and call for creating a policy analysis framework that integrates AI and machine learning and forces deeper interagency coordination.
First, during the transition, the Biden and Trump teams should work together to archive key documents, particularly the Summary of Conclusions (SOCs) and decision memos, that chart key interagency debates around core foreign policy issues. Imagine a team using a large-language model optimized based on SOCs alongside new intelligence estimates using platforms like Maven Smart System. Policy coordination could balance historical insights and recent debates about response options alongside a mix of open-source data and intelligence estimates to recommend new approaches to enduring national security challenges. For example, applied to war in Ukraine, the Trump team could rapidly assess systemic concerns in the Biden administration about escalation and develop new options for positioning Kyiv to negotiate from a position of strength.
This optimized model doesn’t replace the role of experts conducting foreign policy analysis. Instead, it streamlines the process by summarizing past debates and drafting sample questions policy teams can ask different departments and agencies to calibrate a menu of response options. This is not science fiction. A system like this can be built today by vendors with a proven track record of operating on classified networks and tailoring large-language models to different workflows.
Second, the incoming NSC staff can use AI to better connect to the larger framework of strategy plans managed by the Pentagon to deter war and help the United States win in competition. As the Trump team builds optimized models supporting its core issue areas, it can complement the effort with new interagency playbooks. These playbooks would also use the models to generate a menu of policy options. These options, in turn, could “learn” from data provided by each respective agency (e.g., the Department of State and the Department of Treasury) based on current programs and activities to create a common operating picture at the strategic level. Teams, in return, can provide data-driven policy recommendations, not assumptions, and find ways to better integrate different instruments of national power.
Furthermore, this data architecture would create a means of stress-testing the foreign policy agenda. Teams could use AI to generate different scenarios and analyze how best to implement a national security strategy. This functionality would support an iterated approach to tabletop exercises and wargames that support strategic analysis. As a result, the strategy would become more iterative, agile, and responsive to changing conditions.
The next administration will face strong strategic headwinds. From the war in Ukraine to perpetual crises in the Middle East and the growing axis of authoritarianism, foreign policy challenges are compounding and evolving. The Trump administration can use AI to help policy teams navigate this complexity and produce a more practical set of policy options to secure the United States’ interests. This challenge starts with integrating new technology into an old national security system.
Benjamin Jensen is a senior fellow for Futures Lab in the International Security Program at the Center for Strategic and International Studies (CSIS) in Washington, D.C., and the Frank E. Petersen Chair of Emerging Technology and professor of strategic studies at the Marine Corps University School of Advanced Warfighting. The views expressed are his own as a private citizen. Jose M. Macias III is an associate data fellow in the Futures Lab within the International Security Program at CSIS.
Foreign policy in the United States relies on a mix of policy analysis, staffing processes, and bureaucratic procedures that have not changed for decades. The National Security Council (NSC), chartered to guide strategy formation and implementation, has proved impervious to advances in data science and artificial intelligence (AI). Instead, it relies on old-fashioned Microsoft Word documents and PowerPoint attachments to shape critical national security policy debates. This contrasts with the private sector, where analysts integrate dynamic and interactive tools like Notion, Jupyter Notebooks, or R Markdown files and other formats that can take real-time data to clean, transform, visualize, or model outcomes and communicate their meaning for evidence-based decisionmaking. Worse still, the government’s approach to strategy has yet to incorporate advances in AI.
The Trump administration has an opportunity to revolutionize the NSC by integrating AI into policy analysis and using new analytical techniques to create interagency playbooks that cut through the Byzantine federal bureaucracy. Taking this leap should go beyond publishing guidelines on the use of AI in government to building a set of analytical processes that take advantage of emerging technology to sustain the United States’ strategic advantage.
The inner workings of the foreign policy machine start with discussions during the transition that form the basis for a directive published in the first 30 days of a presidency outlining the structure and processes of the NSC. For example, the Memorandum on Renewing the National Security Council System was published on February 4, 2021, 15 days into the Biden administration. Similarly, the first Trump administration published a similar document just 8 days into his administration, on January 28, 2017. Setting the structure and processes early, this system took shape in the late 1980s, under President George H.W. Bush’s National Security Directive 1, published January 30, 1989.
These documents turn the National Security Council Act of 1947 into a bureaucratic reality that sees a symphony of policy coordination and interagency bargaining. Often the process changes are minor, with assistant security–level officials tasked to provide policy analysis for senior leader consideration based on a set of regional and topical issues directed by the NSC Deputies Committee. The structure changes are more significant, with core issues (e.g., competition with China, ending the war in Ukraine and reconstruction, and economic security) creating, in effect, a foreign policy focus and battle rhythm a president’s foreign policy agenda.
Regardless of the core issues around which the Trump team builds the NSC structure, the subsequent “National Security Memorandum 2” should take a bold leap and call for creating a policy analysis framework that integrates AI and machine learning and forces deeper interagency coordination.
First, during the transition, the Biden and Trump teams should work together to archive key documents, particularly the Summary of Conclusions (SOCs) and decision memos, that chart key interagency debates around core foreign policy issues. Imagine a team using a large-language model optimized based on SOCs alongside new intelligence estimates using platforms like Maven Smart System. Policy coordination could balance historical insights and recent debates about response options alongside a mix of open-source data and intelligence estimates to recommend new approaches to enduring national security challenges. For example, applied to war in Ukraine, the Trump team could rapidly assess systemic concerns in the Biden administration about escalation and develop new options for positioning Kyiv to negotiate from a position of strength.
This optimized model doesn’t replace the role of experts conducting foreign policy analysis. Instead, it streamlines the process by summarizing past debates and drafting sample questions policy teams can ask different departments and agencies to calibrate a menu of response options. This is not science fiction. A system like this can be built today by vendors with a proven track record of operating on classified networks and tailoring large-language models to different workflows.
Second, the incoming NSC staff can use AI to better connect to the larger framework of strategy plans managed by the Pentagon to deter war and help the United States win in competition. As the Trump team builds optimized models supporting its core issue areas, it can complement the effort with new interagency playbooks. These playbooks would also use the models to generate a menu of policy options. These options, in turn, could “learn” from data provided by each respective agency (e.g., the Department of State and the Department of Treasury) based on current programs and activities to create a common operating picture at the strategic level. Teams, in return, can provide data-driven policy recommendations, not assumptions, and find ways to better integrate different instruments of national power.
Furthermore, this data architecture would create a means of stress-testing the foreign policy agenda. Teams could use AI to generate different scenarios and analyze how best to implement a national security strategy. This functionality would support an iterated approach to tabletop exercises and wargames that support strategic analysis. As a result, the strategy would become more iterative, agile, and responsive to changing conditions.
The next administration will face strong strategic headwinds. From the war in Ukraine to perpetual crises in the Middle East and the growing axis of authoritarianism, foreign policy challenges are compounding and evolving. The Trump administration can use AI to help policy teams navigate this complexity and produce a more practical set of policy options to secure the United States’ interests. This challenge starts with integrating new technology into an old national security system.
Benjamin Jensen is a senior fellow for Futures Lab in the International Security Program at the Center for Strategic and International Studies (CSIS) in Washington, D.C., and the Frank E. Petersen Chair of Emerging Technology and professor of strategic studies at the Marine Corps University School of Advanced Warfighting. The views expressed are his own as a private citizen. Jose M. Macias III is an associate data fellow in the Futures Lab within the International Security Program at CSIS.
