Study Quantifies the Scientific Cost of U.S. Dual-Use Research Oversight

A large-scale empirical paper published in Science on June 5 puts numbers to a tension that biosecurity policymakers have long acknowledged but rarely quantified: tightening security oversight on dual-use research carries a measurable cost to scientific output, and the controls may not stop the research from happening elsewhere.

The study, "Dual-use research under scrutiny," was authored solely by Professor Seokbeom Kwon of KAIST's School of Business and Technology Management. <cite index="18-8">Kwon developed a methodology combining the U.S. Patent and Trademark Office's multi-stage security review process with patent-paper citation data, then analyzed approximately 600,000 research papers.</cite> The dataset and code have been deposited publicly, per a citation in Science's reference list reviewed by this reporter.

The core empirical finding is straightforward and, for policymakers, uncomfortable. <cite index="21-8,21-9">The analysis showed that dual-use research consistently has greater scientific impact than comparable research, meaning research subject to security oversight tends to play an important role in scientific progress and technological innovation.</cite> That is not a new suspicion; it is now a data-supported claim drawn from a population far larger than any prior case study.

The structural problem Kwon identifies is jurisdictional. <cite index="24-6">Frameworks governing security oversight like National Security Decision Directive 189 (NSDD-189) apply primarily when federal funding is involved, leaving a pronounced regulatory blind spot for research conducted without U.S. governmental direct participation.</cite> That blind spot has grown over time. <cite index="18-12">The share of dual-use research directly involving the U.S. federal government decreased from about 41 percent in 1981 to about 22 percent in 2005.</cite> The paper does not update that figure past 2005, but the direction of the trend lines up with the broad privatization of life-sciences R&D over the same period.

The finding lands in the middle of an active regulatory push. <cite index="17-6">Executive Order 14292, signed in May 2025, intensified federal oversight of biological research with potential security implications, including dangerous gain-of-function research.</cite> That order is also the backdrop for a separate, ongoing CSIS analysis, published two weeks ago, which noted that the FY 2026 budget proposal threatens cuts of $18 billion to NIH, $3.6 billion to CDC, and $240 million to the Administration for Strategic Preparedness and Response, and that the White House Office of Pandemic Preparedness and Response Policy currently has no confirmed leader following the departure of Gerald Parker.

Kwon's argument is not that oversight is wrong. <cite index="28-3,28-4">Although dual-use research warrants oversight given its implications for national security, the regulatory changes prompt critical questions about the benefits and costs of strengthened security regimes, and restrictive measures risk delaying diffusion of crucial scientific discoveries.</cite> The paper's policy prescription is multilateral rather than unilateral: <cite index="20-5">as Kwon stated in a KAIST release, "strengthening security oversight on dual-use research by a single country alone may impose disproportionate costs on domestic science, while having structural limits in preventing the development of equally important research conducted overseas."</cite>

That framing has limits this reporter would flag directly. The study uses patent review and citation data as proxies for dual-use classification. Patents are one slice of the dual-use research universe; a significant share of sensitive life-sciences work, including much gain-of-function research and academic virology, never enters the patent system. The dataset may therefore undercount the problem space, particularly for the biological-weapons-relevant research that biosecurity practitioners most often cite. Confidence in the generalizability of the findings is moderate at best until the methodology is stress-tested against non-patent-bearing datasets.

Still, the paper's core contribution is real. <cite index="18-9">The work has been recognized in academia for shifting discussions of dual-use research, which had previously relied largely on case-based analysis, toward large-scale empirical analysis.</cite> <cite index="14-5">The implications extend beyond biological sciences, as Kwon's study anticipates similar challenges in emerging high-stakes fields such as artificial intelligence and quantum technologies.</cite> That is a dual-use ambiguity that is not resolved by this paper but is at least named honestly.

Sources cited:
- Science (DOI: 10.1126/science.aee2479) (https://www.science.org/doi/10.1126/science.aee2479)
- EurekAlert / KAIST press release (https://www.eurekalert.org/news-releases/1131017)
- Homeland Security Newswire (https://www.homelandsecuritynewswire.com/dr20260608-first-largescale-empirical-analysis-of-dualuse-research-and-security-oversight)
- Bioengineer.org (https://bioengineer.org/kaist-study-delivers-first-comprehensive-large-scale-analysis-of-dual-use-research-and-security-oversight/)
- Seoul Economic Daily (AI translation) (https://en.sedaily.com/technology/2026/06/05/kaist-tougher-research-curbs-over-security-fears-may-stifle)
- CSIS / AI-Enabled Bioterrorism analysis (https://www.csis.org/analysis/opportunities-strengthen-us-biosecurity-ai-enabled-bioterrorism-what-policymakers-should)