Why Human Intelligence Matters More in an AI World

This commentary was originally published by The Cipher Brief on June 17, 2026.

An impending casualty of artificial intelligence, we are told, is the human spy. The conventional wisdom is that in our AI future, there's little need to recruit agents, plan secret rendezvous, or conduct dead drops—the old-school tradecraft of espionage. “Human spies in the field will become rare,” wrote David Ignatius after surveying the growing field of AI intelligence startups. “The future of espionage is written in zeros and ones.”

Yes, AI will become an indispensable tool for case officers, agents, fabricators, counterintelligence services, and the rest of the intelligence world. But this will have the paradoxical effect of increasing the importance of old-fashioned human intelligence.

Economics tells us that the value of something is determined by the benefit it provides on the margin—not by its raw power or pervasiveness. We should expect that AI will lower many barriers to technical intelligence collection and analysis. This has happened in other spheres: Open-source encryption put tools once reserved for states into everyone's hands. Likewise, you no longer need a national space program—just a credit card—to order up high-resolution satellite photography and buy AI-powered software to analyze the results.

Human intelligence is scarcer and harder to replicate, and so its marginal value will rise. Vast amounts of data alone cannot reveal what a foreign leader intends to do, nor what is happening inside an “air-gapped” network. Human sources can.

A human agent can help cut through the deluge of disinformation—and help distinguish authentic source material from forgeries.

Human intelligence has always helped validate technical collection. That role will grow in importance as AI poisons the information environment. Electronic channels will be flooded with fabricated phone calls, forged documents, and other convincing synthetic media. Signals intercepts will be harder to trust on their own. A human agent can help cut through that deluge of disinformation—and help distinguish authentic source material from forgeries.

Finally, AI is making digital communications less trustworthy, as we already see with scammers using deepfake images, audio, and video calls for fraud. As deepfakes get harder to detect, the rational response will be to distrust electronic communications. The corollary? Secure, person-to-person communication will rise in value. Case officers have long relied on dead drops, brush passes, and brief meetings to communicate with their agents. In an AI-saturated world, these traditional techniques may be the most reliable methods of communication we have.

None of this means that technical intelligence will become unimportant. And AI will certainly transform human intelligence operations—for better and for worse. It will help case officers select operational sites, prepare for agent meetings, and improve their ability to persuade foreigners to commit espionage. At the same time, AI will supercharge surveillance, provide fabricators with an unlimited source of plausible but false information, and help counterintelligence services predict the activities of case officers operating in their countries.

Still, as AI makes technical collection cheaper, deception easier, and digital communication less trustworthy, the value of human sources will rise. And the case officer, working in the shadows of our artificial future, will matter more than ever.

Thomas Mulligan is an adjunct researcher at RAND and a visiting scholar at Georgetown University.