The use of artificial intelligence (AI) has the potential to weaken democratic accountability for consequential US national security choices, including war. To ensure that these choices adhere to the public law values of rationality, legality, and accountability, we rely on Congress and the courts, as well as foreign allies, whistleblowers, and even US technology companies; their work, however, is hindered by the classified, 'black box' nature of the Executive's national security decision-making. The Double Black Box: National Security, Artificial Intelligence, and the Struggle for Democratic Accountability addresses these pressing challenges.

The rise of AI systems to enhance national security decision-making - or even make autonomous decisions - deepens this critique, because it is difficult to understand how AI algorithms, often described as 'black boxes,' reach their conclusions. How can we be confident that these AI systems comport with our laws and values? The widespread use of AI inside the national security ecosystem renders US national security choices even more opaque to the public, congressional overseers, US allies, and even the officials making the decisions.

This 'double black box' raises critical challenges for democratic accountability. Because China is committed to becoming the world leader in AI and faces fewer legal and values-based constraints on its pursuit of military AI, democracies' commitment to using AI in lawful and ethical ways will be tested. The book defines and explores this phenomenon and then identifies ways that policymakers, military and intelligence officials, and lawyers in democratic states such as the United States can reap the advantages of advanced technologies without surrendering their public law values.