Treverton brings a rare vantage point to the subject: he chaired the National Intelligence Council, directed research centers at RAND, and has spent decades moving between government service and academic and policy circles. His report treats the shift from scarce, secret information to an abundance of open information as the defining fact of the current era, and makes the case that agencies serve policymakers best when they leverage the full array of available data.
From Scarcity to a Flood of Open Information
During the Cold War, intelligence agencies organized themselves around traditional sources — human spies, intercepted signals, satellite imagery. Today, one CIA veteran estimates that open-source material now makes up two-thirds of the information of genuine interest to U.S. intelligence, and researchers put the volume of available data at more than 180 zettabytes — triple what existed just five years earlier. Crowdsourced investigators, business intelligence firms, and AI startups increasingly work the same ground intelligence agencies once considered exclusively theirs. Open-source teams have on occasion unearthed assessments faster than official channels.
Treverton includes several caveats. Open information still needs verification, including confirmation of its source and context. And high data volume requires sorting signal from noise and takes real resources.
Three Recommendations
Learn from what already worked. The report surveys efforts to bridge government and outside expertise — the NIC’s Intelligence Community Associates, the Pentagon’s Minerva Research Initiative, NSA’s Laboratory of Analytic Sciences at North Carolina State, the University of Maryland’s ARLIS, and In-Q-Tel. These experiences provide the foundations of a working playbook for lasting collaboration among government, academia, and industry.
Treat AI as a partner, with analysts firmly in charge. Treverton envisions AI integration as a three-step evolution: analysts first learn to work alongside AI systems, then apply critical thinking and context to what the systems surface, and finally settle into a role centered on oversight, ethics, and strategic judgment. AI can assemble an assessment in hours that once took a team of analysts months — but hallucinations and pattern-only correlation mean humans stay essential to understanding what the data actually means.
Build structures designed for an open world. The report’s recommends a dedicated Open Source Intelligence Agency, to complement current open source analysis channels. This new agency could serve as a focal point for tradecraft, an incubator for AI-driven methods, and a bridge to academia, Wall Street, and Silicon Valley.
Intelligence as a Shared Resource
The report speaks to a wide readership by design — intelligence executives navigating the boundary between classified and open sources, policymakers who depend on timely and trustworthy analysis, and the broader community of scholars and practitioners thinking through how institutions adapt in the age of AI and similar advances in the information landscape.
