Biosecurity: 6% of Countries Have Any Oversight, There Is No US Federal Law, and the Screening Requirement Was Ordered in 2025 and Dropped.

In short
Key takeaways
- No single US federal law governs lab biosafety with enforceable penalties; oversight is a mixture of federal law, guidance, and self-governance
- Only 6% of countries have national-level oversight measures for dual-use bioscience research, per the Global Health Security Index
- Executive Order 14292 mandated two policy revisions in May 2025 with 90 and 120 day deadlines; one was delivered in July 2026, the other was not
- The Federal Select Agent Program caps emergency waivers at 60 days by statute, and HHS has not assessed the risk this poses to oversight of pandemic pathogens
- A GAO-23-105455 finding says HHS provides no standard for what "reasonably anticipated" means, leaving interpretations subjective and inconsistent
The reason the 2024 to 2026 gain-of-function debate produced no regulation is not that the science is unclear. It is that there is no institution to regulate it.
Start with the facts that are not in dispute, because they come from the government's own research arm. The Congressional Research Service: "In the United States, there currently is no single, overarching federal law governing biosafety and biosecurity with enforceable legal penalties beyond those in the Federal Select Agent Program." Oversight is "a mixture of federal law, federal guidance, and self-governance," dependent on the types of experiments and agents being used. Per the Global Health Security Index, only 6% of countries have national-level oversight measures for dual-use bioscience research. There is no international entity whose primary mission is reducing emerging biological risks. A witness testifying to Congress asked for "an independent biosafety and biosecurity oversight authority, analogous to the FAA's oversight over air transportation," and it does not exist.
And the detail that makes this concrete rather than abstract: Executive Order 14292, signed 5 May 2025, directed OSTP to revise or replace two policies. The White House delivered one in July 2026 and never delivered the other. Seventeen months, one of two.
Four partial regimes in a trench coat
CRS document R47114 describes the overlap problem precisely: "Overlapping policies and guidance address aspects of biosafety and biosecurity associated with GOF research in the United States - some impose requirements, some provide guidelines, some apply only to research with select biological agents, and some apply only to federally funded research." Four different enforcement scopes, applied to the same activity. That is not regulation. It is four partial regimes with gaps between them, and the gaps are the governing document.
How the gap gets filled anyway
A 2025 congressional hearing put the shape of it: "Over the past 20 years, a lot of institutions have adopted the same committees, roles, biocontainment categories, and reporting structures for virtually all life sciences research, regardless of funding source or discipline. This shared structure has allowed a fragmented regulatory environment to function as a system."
Read that sentence again with the emphasis on the second half. The biosafety committee at your local university exists, does real work, and covers most of the research. It works because a coincidental convergence of institutional practice filled a gap that law left open. And the Bulletin of the Atomic Scientists names the actual load-bearing material: "The guidelines are tied to NIH funding, institutions maintain the programs necessary to remain eligible for NIH grants."
US biosecurity governance is implemented by grant terms, not by law. That is a real system, it functions today, and it is one appropriations decision away from not existing. The same conditionality that makes the synthesis screening framework partial also underwrites the entire domestic compliance architecture. Checking EO 14292 line by line against its own deadlines shows whether the grant-term architecture was being converted into law. The answer is in the next section.
The compliance table
The table below was assembled for this post, because it did not exist anywhere, and it is entirely checkable from CRS, Federal Register and OSTP sources:
| Directed by EO 14292, May 2025 | Deadline | Delivered |
|---|---|---|
| Revise or replace the 2024 policy for Dual Use Research of Concern and Pathogens with Enhanced Pandemic Potential | 120 days | Yes, replaced July 2026 by the US Government Policy for Stopping High-Risk Life Sciences Research |
| Revise or replace the 2024 Framework for Nucleic Acid Synthesis Screening | 90 days | No |
The Congressional Research Service records the second one plainly: the 2026 policy "did not revise or replace the 2024 Framework for Nucleic Acid Synthesis Screening." The 90-day directive has been outstanding for 17 months.
Why does the unmet one matter more than the met one? Because nucleic acid synthesis screening is the single chokepoint in the entire biosecurity architecture. You cannot build a pathogen you cannot order the sequence for. It is upstream of every other control. The 2024 framework covers recipients of federal funding, and it is a funding condition, not a legal requirement, so it does not reach researchers without federal money. Congressional testimony stated the requirement for what has been needed since 2021: "establishing this as a legal requirement will also be important." It still is not one. EO 14292's own words asked for screening mechanisms that are "comprehensive, scalable, and verifiable." What exists is a funding condition with a 90-day replacement deadline that came and went more than a year ago.
The 60-day statutory cap, which is the detail people remember
The GAO finding in GAO-23-105455 cuts in two directions, and both are unhinged. First: HHS's Framework "does not fully meet the key elements of effective oversight" because it "does not provide a standard to help funding agencies interpret what 'reasonably anticipated' means," which leaves "subjective and potentially inconsistent interpretations" and leaves HHS "without assurance the department is reviewing all necessary research proposals."
Second, and this is the finding that is hardest to read: HHS faces trade-offs in adding newly emerged pathogens, like SARS-CoV-2, to the Federal Select Agent list because, as officials told GAO, doing so "would impede the public health response by burdening diagnostic and treatment facilities with additional reporting and inspection requirements." And "the statute authorizing the Federal Select Agent Program limits HHS's ability to waive or postpone these requirements during public health emergencies for a maximum of 60 days. HHS has not assessed the risk this limitation poses to its oversight of known pandemic pathogens."
The mechanism designed to make pathogens visible to regulators is structurally suppressed during exactly the circumstances when a new pandemic pathogen appears. For up to 60 days, by statute. And HHS has not assessed the risk. HHS and OIG reporting contains no follow-up assessment. That is a GAO finding with no follow-up, and and it is the line least likely to appear in coverage of the BIOSECURE Act.
The international picture, and the FAA that does not exist
Testimony to Congress puts it in one breath: "biosafety and biosecurity are very weak globally," only 6% of countries have national-level oversight measures for dual-use bioscience research, and there is "no international entity that has its primary mission dedicated to reducing emerging biological risks."
The witness's recommendations were concrete: adopt and implement policy responsibly, authorise and fund an independent biosafety and biosecurity oversight authority analogous to the FAA, establish a legal requirement to screen DNA synthesis orders, implement existing strategy documents, and do the diplomatic work. The FAA analogy is the whole argument. Aviation has an independent authority with enforcement power, a licence regime, and a mandated reporting culture, and it took thirty years of institutional building. Biosecurity has a committee structure that a researcher maintains. The chairman of the relevant subcommittee added the geopolitical asymmetry himself: "Not every country - or every company operating within it - shares the same concerns about biosafety and biosecurity."
The BIOSECURE Act is not what the coverage says it is
The BIOSECURE Act was enacted in December 2025, and it is routinely described as a biosecurity law. It is not one. From the subcommittee chairman's own opening statement, it "establishes restrictions on federal procurement and contracting involving equipment, services, or other capabilities provided by identified biotechnology companies." That is a supply-chain restriction on federal purchasing, a real tool aimed at who sells to the US government. The legislative response to two years of gain-of-function debate is a procurement exclusion list. It is also not the regulation that does not exist, and the coverage that lets the enactment stand in for the missing law is doing exactly the substitution the agencies have been doing.
AI moved the capability curve, and the institutions did not move at all
The Frontiers in Microbiology paper from Kritika Mohan et al., "Know your scientist: KYC as biosecurity infrastructure," maps the new capability curve. AI-assisted biological design tools now let researchers generate candidate designs. But the paper's own assessment: "Functional annotation remains incomplete, and reliable prediction of pathogenicity or toxicity is beyond current capabilities." And "Most biological design tools require downstream wet-lab validation to have any practical impact, which in turn requires institutional affiliation or formal collaboration."
That second sentence is the most important technical fact in this whole story. The institution is already the chokepoint. If wet-lab validation requires institutional affiliation, then institutional vetting is the correct primary control, which is why the funding-conditioned architecture matters so much. The paper proposes three tiers: Tier I makes research institutions trust anchors vouching for affiliated researchers; Tier II does output screening via sequence homology search and functional annotation; Tier III does behavioural monitoring for anomalies inconsistent with declared research purposes. And the limits the paper states: annotation is incomplete, pathogenicity prediction is beyond current capability, so Tier II screening will produce false positives that land on exactly the legitimate researchers the system is supposed to protect, and governing only at the synthesis stage "creates gaps" because "researchers can accumulate dangerous knowledge, share designs through alternative channels, or synthesize abroad."
Meanwhile, the pipeline is moving the other way. S.4770, the Web of Biological Data Act, would centralise biological data for AI research. H.R.7801, S.2676 and S.3468 would direct NSF to establish a national network of autonomous and self-driving labs. The plain question is what the bills actually say and who would license a self-driving lab: under what screening regime would autonomous lab capacity be permitted to run? The answer from the bill text is that no existing regime covers it. A national self-driving lab network is a scaling decision about dual-use capability being made in an appropriations bill, and the controls that would need to scale with it are the ones that do not exist. It cuts both ways: more throughput for legitimate research, and equally more throughput for anything the screening misses.
Where this could be wrong
Four objections deserve a direct answer. First, this plays into a political agenda; the substance here is documentary, a CRS statement, a GAO finding, and a CRS record of an unmet directive, none of it partisan, all of it checkable from institutional sources, and the BIOSECURE Act and the delivered 2026 policy deserve to be counted as real achievements where they are real. Second, the biology is the dual-use problem, not the paperwork. The KYC paper makes that point: the point being that the capability curve and the control curve are moving at different speeds and the control curve is conditioned on funding. Third, gain-of-function research has been regulated for decades and came out fine; the framework is legitimate and also lacks a floor: the 2017 DURC policy only reaches federally funded research, GAO found no standard for "reasonably anticipated," and the 60-day cap is statutory. A framework with no floor and no enforcement authority is not a framework; that is a procedural claim, not a conspiratorial one. Fourth, the 6% figure may understate partial coverage, and the stronger claims are the domestic ones. Fifth, this is a US problem, not a global one, except the Global Health Security Index makes it global, and the highest-burden pathogen exposures are not in the countries with the oversight.
Biosecurity has a committee, not an authority
The claim is narrower than the headline suggests, and worth stating exactly. The committees do real work. The 2024 policies are real. The BIOSECURE Act is a real tool. Congress is actively hearing these cases. The claim is about the architecture: conditional, funding-linked, and without an enforcement authority, and about where accountability sits if the committees fail. The answer is nowhere. The screening framework that sits upstream of every other control was ordered to be replaced within 90 days in May 2025, and 17 months later it is still the old one, still a funding condition, still not a law. Biosecurity got an executive order and a committee. Aviation got an FAA. Read those two sentences back to back and the comparison does the work.
On this page
- Four partial regimes in a trench coat
- How the gap gets filled anyway
- The compliance table
- The 60-day statutory cap, which is the detail people remember
- The international picture, and the FAA that does not exist
- The BIOSECURE Act is not what the coverage says it is
- AI moved the capability curve, and the institutions did not move at all
- Where this could be wrong
- Biosecurity has a committee, not an authority
Sources
- Artificial Intelligence and Biosecurity Issues (CRS IF13269)Congressional Research Service, 2026
- Public Health Preparedness: HHS Could Improve Oversight of Research Involving Enhanced Potential Pandemic Pathogens (GAO-23-105455)US Government Accountability Office, 2023
- Strengthening Biosafety and Biosecurity Standards hearing recordUS House of Representatives, 2022
- Know your scientist: KYC as biosecurity infrastructureMohan K et al., Frontiers in Microbiology, 2026
- NIH modernization could expose biosafety gapsBulletin of the Atomic Scientists, 2026
Frequently asked questions
Is there actually no US law on lab biosafety?
There is no single overarching federal law governing biosafety and biosecurity with enforceable legal penalties. Oversight is a mixture of federal law, federal guidance, and self-governance depending on the experiment and agent. The Federal Select Agent Program is the closest thing to binding law, a list-based programme covering certain agents and toxins. The GAO found in 2023 that HHS oversight of enhanced pandemic pathogen research does not meet key elements of effective oversight because the Framework provides no standard for what reasonably anticipated means.
What did Executive Order 14292 do and not do?
It directed OSTP to revise or replace the 2024 dual use research oversight policy within 120 days, and the 2024 Framework for Nucleic Acid Synthesis Screening within 90 days. In July 2026 the White House released the US Government Policy for Stopping High-Risk Life Sciences Research, replacing the first policy. The Congressional Research Service records that the order did not revise or replace the screening framework, so one of two mandated changes was delivered 17 months later.
Why is nucleic acid synthesis screening the chokepoint?
The 2024 framework established screening requirements for recipients of federal funding for synthetic nucleic acid research. You cannot build a pathogen you cannot order the sequence for, so it sits upstream of every other control. But it is conditioned on federal funding rather than being a legal requirement, it does not reach researchers without federal money, and congressional testimony has called for making it a legal requirement, which it still is not.