The Pentagon nuclear command, control, and communications enterprise is decades old and desperate for an upgrade, says the head of U.S. Strategic Command, and artificial intelligence could help fortify nuclear C3 for its no-fail mission.
STRATCOM is “exploring all possible technologies, techniques, and methods to assist with the modernization of our NC3 capabilities,” said Air Force Gen. Anthony J. Cotton, who has led the command since December 2022.
“AI will enhance our decision-making capabilities,” Cotton said at the 2024 Department of Defense Intelligence Information System Conference. “But we must never allow artificial intelligence to make those decisions for us.”
Growing threats, an overwhelming flow of sensor data, and increasing cybersecurity concerns are driving the need for AI to keep American forces a step ahead of those seeking to challenge the U.S., Cotton said. “Advanced systems can inform us faster and more efficiently,” he explained. “But we must always maintain a human decision in the loop to maximize the adoption of these capabilities and maintain our edge over our adversaries.”
Cotton said AI can help give leaders more “decision space” to ensure the entire nuclear enterprise stays secure. “Our adversaries must know that our nuclear command and control and other capabilities that provide decision advantage are at the ready, 24/7, 365 and cannot be compromised or defeated,” Cotton said.
Cotton’s predecessors at STRATCOM, Adm. Charles Richard and Gen. John E. Hyten, also addressed NC3 modernization. But at a time when the Air Force is also trying to modernize its strategic bomber force with the B-21 Raider bomber, the ICBM force with the Sentinel missile, and its ballistic missile submarine force with the Columbia class submarines, NC3 gets little attention—even though none of those systems can be effective without it.
“Despite warnings from top national security officials, important improvements to NC3 have been fragmented,” wrote Peter L. Hays of the Center for Strategic and International Studies and consultant Sarah Mineiro in an Oct. 28 blog post for the Atlantic Council.
Heather Penney, senior resident fellow at AFA’s Mitchell Institute for Aerospace Studies, noted in a recent podcast that NC3 is often taken for granted, “because it’s largely invisible … underground cables, computers, communications links, and a very few specialized aircraft and satellites are the backbone of this mission function,” she said. “But it’s not like we see those things at air shows or on promotional posters.”
Chris Adams, general manager of Northrop Grumman’s Strategic Space Systems Division, said the real challenge with NC3 is that it’s comprised of so many pieces. “It’s not a single system that was deployed at one point in time,” he said on the Mitchell Institute podcast. “It’s a system of systems. Itreally encompasses hundreds of individual systems that are modernized and sustained over a long period of time in response to an ever-changing threat.”
Injecting AI into some of those systems offers the possibility of more speed and the ability to make sense of huge amounts of information drawn in by that system of systems. There are risks with AI, as researchers have noted. including misplaced trust in AI, “poisoned” data ingested into systems, inaccurate algorithms, and more.
Cotton accounted for such pitfalls, but sees greater promise overall. “Advanced AI and robust data analytics capabilities provide decision advantage and improve our deterrence posture,” he said. “IT and AI superiority allows for more effective integration of conventional and nuclear capabilities, strengthening deterrence.”
AI could be used to automate data collection and accelerate data sharing and integration with allies, he suggested. But he also said “we need to direct research efforts to understand the risks of cascading effects of AI models, emergent and unexpected behaviors, and indirect integration of AI into nuclear decision-making processes.”
Implementing new NC3 systems—with and without artificial intelligence—will need to be a deliberate process said, Adams said.
“We have to consider when, where, and how we want to deploy the next generation of systems incrementally and carefully, so we don’t leave any vulnerabilities,” Adams said. “A good analogy is grabbing the next ring on the playground before letting go of the last one.”