Science Study Quantifies the Cost of Single-Country Oversight on Dual-Use Research

A peer-reviewed paper published in Science on June 5, 2026 puts the hardest empirical numbers yet on a tension that biosecurity policymakers have long talked around: unilateral oversight of dual-use research can throttle domestic science without meaningfully slowing the global research frontier.

The study, authored by Professor Seokbeom Kwon of KAIST's School of Business and Technology Management, drew on roughly 600,000 research papers, combining patent and publication data to map where dual-use research of concern (DURC) gets done, how it flows across borders, and what happens to scientific output when one country tightens controls. According to the KAIST press release reviewed by EurekAlert, the analysis found structural limits to what any single nation's ex-ante oversight regime can accomplish.

The timing is pointed. Executive Order 14292, signed in May 2025, intensified federal oversight of biological research with security implications, including gain-of-function studies, according to the EurekAlert release. That order represents the kind of unilateral tightening Kwon's data now explicitly interrogates. The paper's core finding, as described in the KAIST announcement, is that strengthening security oversight 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.

That framing deserves a careful read. It's not a claim that oversight is pointless. It's a claim that the cost-benefit calculus looks very different when you account for geographic substitution: restrict a line of research in the U.S., and the work may simply migrate to jurisdictions with lighter controls. The paper's policy implication, according to the Seoul Economic Daily's coverage of the release, is that balanced policies between scientific progress and national security are necessary, and that international cooperation is the instrument most likely to reduce structural tension rather than heighten it.

For anyone tracking the DURC debate in Washington right now, the paper lands in a complicated environment. The Trump administration's FY 2026 budget proposed cuts of roughly $18 billion to NIH and $3.6 billion to CDC, according to a CSIS analysis reviewed separately. The White House Office of Pandemic Preparedness and Response Policy is operating without a permanent director following the recent resignation of Gerald Parker, as CSIS noted in the same document. Against that backdrop, the argument that the U.S. needs more international coordination on DURC oversight, not less, is going to find a skeptical audience in parts of the executive branch even as Kwon's data push in that direction.

The paper also enters the field just as OpenAI's Rosalind Biodefense program, launched May 29, 2026, is routing a frontier life-sciences model to vetted government and research partners. According to Tech Times's coverage of the launch, OpenAI's own CEO joined other AI company leaders in warning Congress on June 4 that AI is eroding the knowledge barriers to bioweapons. That warning and Kwon's paper are pointing at the same underlying problem from different angles: the dual-use surface area of biology is expanding faster than oversight frameworks can track it, and the frameworks that do exist are calibrated for a pre-AI research environment.

Kwon's paper is the right kind of contribution for that moment. It doesn't resolve the dual-use dilemma and doesn't pretend to. What it does is replace an argument that has mostly run on assertion with actual citation and publication data at scale. Policymakers who want to justify tighter unilateral controls now have to engage that evidence base. Those who want to justify doing nothing have the same problem. The paper, titled "Dual-use research under scrutiny" (DOI: 10.1126/science.aee2479), is the place to start.

Sources cited:
- EurekAlert (KAIST press release) (https://www.eurekalert.org/news-releases/1131017)
- Science (journal, DOI: 10.1126/science.aee2479) (https://www.science.org/doi/10.1126/science.aee2479)
- Seoul Economic Daily (https://en.sedaily.com/technology/2026/06/05/kaist-tougher-research-curbs-over-security-fears-may-stifle)
- CSIS, Opportunities to Strengthen U.S. Biosecurity from AI-Enabled Bioterrorism (https://www.csis.org/analysis/opportunities-strengthen-us-biosecurity-ai-enabled-bioterrorism-what-policymakers-should)
- Tech Times, OpenAI's Rosalind Biodefense Opens GPT-Rosalind to Vetted Partners (https://www.techtimes.com/articles/318356/20260614/openais-rosalind-biodefense-opens-gpt-rosalind-vetted-partnersdual-use-fears-mount.htm)