Gain-of-function research is scientific work that deliberately alters an organism, usually a virus, to give it new or enhanced abilities such as increased transmissibility, expanded host range, or greater resistance to drugs and vaccines. Scientists do it primarily to understand how dangerous pathogens might evolve in nature, to build animal models of human disease, and to get ahead of potential pandemics by developing vaccines and treatments before a worst-case outbreak materializes. The practice sits at one of the sharpest tension points in modern science: the same experiments that could help prevent a pandemic could, if something goes wrong, start one.
What Scientists Actually Do in These Experiments
The term “gain of function” is broad enough to cover routine lab work, like tweaking a virus so it glows under a microscope, all the way up to engineering changes that make a pathogen more lethal or contagious. The experiments that stir public concern are at the dangerous end of that spectrum. In those cases, researchers typically modify a virus through targeted genetic changes, serial passage through animal hosts, or both, aiming to enhance specific traits: how easily the virus spreads through the air, how quickly it replicates in cells, how effectively it dodges immune defenses, or how well it infects species it normally cannot.
These enhanced traits are not goals in themselves. They are tools for answering questions that are otherwise unanswerable. If you want to know which mutations would allow a bird flu virus to spread between people, you cannot simply wait for nature to run the experiment. You can, however, introduce targeted mutations in a lab and observe what happens in an animal model. The results tell you which genetic signatures to watch for in circulating strains, where the real-world danger thresholds lie, and what kind of countermeasures might work.
Why Researchers Say It Is Worth the Risk
The scientific case for gain-of-function work rests on several pillars, each addressing a gap that other research methods struggle to fill.
The first is pandemic preparedness. By studying how a virus could become more dangerous, researchers aim to identify warning signs before a natural outbreak catches the world off guard. If scientists already know that a particular set of mutations in an avian influenza virus would allow airborne spread among mammals, surveillance networks can watch for those mutations in birds and pigs circulating on farms. This idea of getting a “head start on a virus” is central to the argument: gain-of-function research can inform vaccine or therapeutic development before a pathogen emerges in its truly virulent form.
The second pillar is animal-model development. Many emerging diseases lack good animal models that mimic how the infection behaves in humans. When a virus does not naturally infect mice or ferrets in a way that resembles human illness, researchers sometimes need to adapt it, introducing mutations that let it replicate in a new host. That process inherently changes the pathogen, and those changes can count as gain-of-function modifications. But without workable animal models, testing new drugs and vaccines stalls.
The third is basic virology. Understanding the molecular mechanisms behind transmissibility, immune evasion, and virulence means researchers can design better-targeted drugs and broader-spectrum vaccines. Passive immunotherapy, which often involves combinations of antibody products, is one area that relies heavily on gain-of-function experiments to evaluate whether treatments actually work against enhanced strains.
The Experiments That Ignited the Debate
Two lines of research, both involving influenza, dragged gain-of-function work from an obscure corner of virology into public consciousness.
The first involved the H5N1 avian influenza virus. H5N1 kills roughly half the people it infects, but it does not spread easily between humans. In 2012, a team led by Ron Fouchier in the Netherlands reported that they had genetically modified H5N1 and then serially passaged it through ferrets, the standard mammalian model for human flu. Over successive rounds of infection, the virus acquired mutations that made it airborne-transmissible between ferrets.
The implications were stark. The experiment demonstrated that H5N1 could, in principle, gain the ability to spread through the air among mammals with only a handful of mutations. The scientific value was real: it identified the specific genetic changes that would signal a dangerous shift in circulating bird flu strains. But the experiment also created, in a laboratory, a version of one of the world’s deadliest viruses that could now spread through respiratory droplets. The resulting uproar over whether such research should be published, let alone conducted, shaped biosecurity policy for a decade.
The second landmark involved the 1918 “Spanish” influenza virus. Using fragments of viral RNA recovered from preserved tissue samples and victims buried in permafrost, researchers reconstructed the complete 1918 virus. When tested using reverse genetics, constructs containing the 1918 surface proteins on a modern flu backbone were lethal in mice, and the fully reconstructed 1918 virus was even more virulent. Nearly a century after the pandemic that killed tens of millions of people, the reconstruction yielded novel insights into flu biology and pathogenesis and provided information relevant to preventing future pandemics. Critics countered that resurrecting a virus responsible for one of history’s deadliest outbreaks was an extraordinary gamble, no matter how robust the containment.
These experiments remain touchstones in the debate. Research on H5N1 in ferrets continues today. A recent study comparing current clade B3.13 H5N1 viruses, the strains circulating in US dairy cattle, with older strains found that none transmitted through the air in ferrets, but the B3.13 viruses shed at significantly higher levels in airways than earlier strains. That kind of finding underscores why surveillance-oriented gain-of-function work persists: the virus is evolving, and researchers want to know how close it is to a dangerous threshold.
How Dangerous Pathogens Are Contained
Gain-of-function experiments on potential pandemic pathogens take place in high-containment laboratories, typically rated at biosafety level 3 (BSL-3) or BSL-4. BSL-3 labs are designed for agents that can cause serious or lethal disease through inhalation. They require negative air pressure so that air flows inward if a door opens, HEPA filtration on all exhaust air, rigorous personnel training, and strict decontamination protocols. BSL-4 facilities add another layer, often requiring researchers to work in full pressure suits with independent air supplies, because the pathogens handled there have no available vaccines or treatments.
Modern high-containment facilities increasingly adopt formal biorisk management systems. These go beyond physical engineering controls to include systematic risk assessment, personnel reliability programs, incident-reporting structures, and regular audits. A well-documented example is the BSL-3 core facility at the National University of Singapore, which implemented a management-system approach specifically to handle multiple risk-group-3 agents including SARS-CoV-2, integrating biosafety and biosecurity precautions into a single framework.
The engineering is impressive, but it is not foolproof. Every layer of protection depends on human compliance, and humans make mistakes.
When Containment Breaks Down
Laboratory-acquired infections are rare on a per-researcher basis, but they are not rare enough to ignore when the pathogen in question could spark an epidemic. A systematic review covering reports worldwide identified 712 human cases of laboratory-acquired infections, with research laboratories accounting for 276 infections and eight deaths, and clinical laboratories accounting for 227 infections and five deaths. The leading risk factors were needlestick injuries, failure to properly use personal protective equipment, and inadequate decontamination techniques.
A separate review focusing on the period between 2000 and 2021 identified 309 individuals who contracted lab-acquired infections across 94 reported incidents involving 51 different pathogens. Eight of those infections were fatal, caused by bacteria like plague and meningococcus, and by Ebola virus. Beyond individual infections, the review documented 16 instances where pathogens actually escaped the laboratory into the surrounding community. Those escapes involved some of the most consequential agents in biosecurity: anthrax, SARS-CoV, and poliovirus each accounted for multiple events.
These numbers come with a major caveat: reporting is inconsistent. Many countries have no formal system for tracking lab accidents, and institutions have obvious incentives to keep incidents quiet. The true frequency of laboratory-acquired infections and pathogen escapes is almost certainly higher than published reports suggest. Modeling work has explored what would happen in the early days of a pandemic sparked by a lab leak from a BSL-3 or BSL-4 facility versus one seeded by a naturally emerging pathogen. One analysis found that the United States, India, and China face the highest projected impact in both scenarios by day 30, driven by population size and global travel connectivity. The model suggests that countries with large numbers of high-containment labs start at an elevated baseline of lab-leak risk simply by virtue of having more facilities.
The US Policy Rollercoaster
The American government’s approach to gain-of-function oversight has swung between alarm and accommodation. In October 2014, the US government imposed a funding pause on gain-of-function research involving influenza, MERS, and SARS viruses, halting roughly two dozen federally funded projects. The moratorium was a direct response to the H5N1 ferret controversy and to a string of embarrassing biosafety lapses at federal labs, including a mishandled shipment of live anthrax spores.
The pause lasted about three years. In December 2017, the government lifted it and replaced it with a new review framework. The Department of Health and Human Services adopted what became known as the P3CO Framework, short for Potential Pandemic Pathogen Care and Oversight. Under this system, proposed research that could enhance the transmissibility or lethality of a potential pandemic pathogen would be evaluated on a case-by-case basis by an independent review board before funding could proceed.
The P3CO Framework was a compromise. Proponents of gain-of-function research welcomed the restoration of funding. Critics argued that the review process lacked transparency and that the criteria for triggering review were too narrow, allowing some risky work to proceed without scrutiny. The framework was updated again in 2024, but the underlying logic has remained consistent across administrations: the benefits of research involving pathogens with enhanced pandemic potential must justify the risks.
Who Decides What Is Acceptable Risk
The ethical questions around gain-of-function research do not reduce to a simple risk-benefit calculation, though that framing gets used a lot. A white paper commissioned by the National Institutes of Health laid out a more nuanced framework, arguing that any given gain-of-function study falls somewhere on an ethical spectrum rather than on one side of a bright line. The framework identifies several dimensions that matter: whether the research addresses a genuine scientific imperative, whether the risks are proportional to the expected benefits, whether risks have been minimized to the extent possible, whether remaining risks are manageable, whether the burdens and benefits are distributed justly, and whether governance is transparent and democratic.
Other ethicists have pushed the argument further. One influential paper contended that bioethical principles normally reserved for human-subjects research, like the Nuremberg Code’s requirement that experiments produce results “unprocurable by other methods” and maintain proportionality between risk and humanitarian benefit, should extend to gain-of-function studies even though the research technically involves no human subjects. The reasoning is straightforward: if an experiment could endanger public health on a massive scale, the ethical standards should reflect that potential, not the fact that the immediate subjects are ferrets or cell cultures.
This is where the debate gets genuinely difficult. Nobody disputes that understanding pandemic-potential pathogens matters. The disagreement is about whether any institutional review process can reliably weigh a speculative future benefit (a vaccine developed faster because of gain-of-function data) against a speculative catastrophic risk (a lab accident that seeds a pandemic). Both sides are arguing about probabilities that are hard to estimate and consequences that are almost impossible to model.
Dual Use and the Biosecurity Dimension
Beyond lab safety, gain-of-function research raises a separate category of concern: dual use. The same knowledge that helps a vaccine developer understand how a virus might evolve can, in theory, help someone deliberately engineer a biological weapon. This is not a hypothetical worry confined to thriller novels. Any published gain-of-function study that identifies the specific mutations needed to make a pathogen more dangerous is, in effect, a partial blueprint.
The concept of dual-use research of concern has been formalized in policy. Some researchers and policy analysts argue that all gain-of-function work should be classified as dual-use research of concern by default, triggering enhanced biosafety controls, regulatory oversight, and engagement with international law. The argument is that the current approach, which classifies studies as dual-use only after specific criteria are met, leaves gaps that could be exploited or overlooked.
Dual-use risks are not distributed evenly around the world. Regions with political instability or weaker biosafety infrastructure face particular challenges. A study examining dual-use research governance in the Middle East found significant variation in how effectively countries in the region implement biosafety and biosecurity protocols compared to global standards. When the expertise and materials for dangerous research spread faster than the oversight frameworks needed to manage them, the global risk profile changes regardless of what any single country’s review board decides.
What “Gain of Function” Does Not Mean
Public discussion of gain-of-function research often collapses the entire category into its most extreme examples, which creates confusion. Most genetic modifications to pathogens in labs are routine and unremarkable. Researchers insert fluorescent markers into viruses to track them under a microscope. They delete genes to see what stops working. They swap protein segments between related strains to map which regions control which functions. All of these technically involve altering an organism’s genetic makeup, and many could be called “gain of function” if you define the term loosely enough, but they pose no pandemic risk.
The experiments that matter for public safety are a narrow subset: those that enhance transmissibility, virulence, or immune evasion in pathogens that could plausibly cause a pandemic. The US policy framework reflects this distinction by focusing specifically on “enhanced potential pandemic pathogens,” a category that filters out the vast majority of lab work involving microbes. When you read a headline about gain-of-function research, it is worth asking which kind the story is actually about, because the risk calculus is completely different for a fluorescent-tagged lab strain and an airborne-transmissible bird flu virus.
Ferret Studies and the Question of Predictive Value
Ferrets occupy a unique place in influenza research because their respiratory tracts resemble ours closely enough to make them useful stand-ins for studying how flu viruses spread. The H5N1 airborne-transmission experiments relied on ferrets for exactly this reason. But a lingering question in the field is how well ferret results actually predict what would happen in humans.
Recent work has tried to sharpen the predictive value of ferret studies by correlating multiple metrics: viral loads in nasal washes, the amount of virus shed into the air, and actual transmission between animals. A 2025 study examining current H5N1 strains found strong correlations between these measures, suggesting that airborne virus shedding in ferrets can serve as a meaningful early indicator of a strain’s pandemic potential. None of the H5N1 strains tested in that study transmitted through the air, but the B3.13 clade viruses produced significantly higher airborne viral loads than older strains. The researchers argued that these metrics could help identify zoonotic influenza viruses that are adapting toward increased transmission potential before they cross the threshold into full airborne spread.
This kind of incremental work rarely makes headlines, but it represents the ongoing, less dramatic side of gain-of-function-adjacent research: refining the tools scientists use to assess risk, so that the surveillance systems watching for the next pandemic have better information to act on.
The Transparency Problem
One of the most persistent criticisms of gain-of-function research governance is that oversight happens largely behind closed doors. In the United States, the P3CO review process does not publish its deliberations, and for years it was unclear how many proposals had even been reviewed, let alone how many had been approved, modified, or rejected. This opacity makes it difficult for the public, and even for other scientists, to assess whether the system is working.
Transparency matters for a practical reason beyond democratic accountability. If the scientific community cannot see which experiments have been reviewed and under what reasoning, it cannot build shared norms about what is acceptable. Individual institutions end up making judgment calls in isolation, and the line between “acceptable gain-of-function research” and “too risky” shifts depending on who is drawing it. International coordination becomes nearly impossible when different countries do not even share a common vocabulary for categorizing risk, let alone common standards for managing it.
Proposals to address this range from mandatory public disclosure of funded gain-of-function projects (with sensitive details redacted) to the creation of an international body modeled on nuclear nonproliferation frameworks. None has gained enough traction to become policy. In the meantime, the gap between the speed at which virology advances and the speed at which governance catches up continues to widen. The technology for manipulating viruses gets cheaper and more accessible every year. The institutions for overseeing that technology remain, in many parts of the world, largely voluntary.