By Kevin Esvelt | November 15, 2022
Pandemics can begin in many ways. A wild animal could infect a hunter, or a farm animal might spread a pathogen to a market worker. Researchers in a lab or in the field could be exposed to viruses and unwittingly pass them to others. Natural spillovers and accidents have been responsible for every historical plague, each of which spread from a single individual to afflict much of humanity. But the devastation from past outbreaks pales in comparison to the catastrophic harm that could be inflicted by malicious individuals intent on causing new pandemics.
Thousands of people can now assemble infectious viruses from a genome sequence and commercially available synthetic DNA, and numerous projects aim to find and publicly identify new viruses that could cause pandemics by characterizing their growth, transmission, and immune evasion capabilities in the laboratory. Once these projects succeed, the world will face a significant new threat: If a single terrorist with the necessary skills were to release a new virus equivalent to SARS-CoV-2, which has claimed 20 million lives worldwide, that person would have killed more people than if they were to detonate a nuclear warhead in a dense city. If they were to release numerous such viruses across multiple travel hubs, the resulting pandemics could not plausibly be contained, and would spread much faster than even the most rapidly produced biomedical countermeasures. And if one of those viruses spread as easily as the omicron variant—which rapidly infected millions of people within weeks of being identified—but had the lethality of smallpox, which killed about 30 percent of those infected, the subsequent loss of essential workers could trigger the collapse of food, water, and power distribution networks—and with them, societies.
To avoid this future, societies need to rethink how they can delay pandemic proliferation, detect all exponentially growing biological threats, and defend humanity by preventing infections. A comprehensive set of directions detailing how we can build a world free from catastrophic biological threats is required. That roadmap now exists.
The threat. Each year, universities and labs around the world train more people in laboratory skills important for biomedical research and the bioeconomy, but many of those same skills can also be used to assemble viruses. Judging from records of doctoral degrees awarded in different fields that acquire the relevant skills, at least 30,000 individuals can successfully follow public step-by-step protocols to obtain any influenza virus with a published genome sequence from commercially available synthetic DNA. Coronaviruses and paramyxoviruses such as MERS and measles require synthetic genes to be assembled into larger genomes, likely cutting the number of people with the requisite skills and resources to the single-digit thousands. Only one or two hundred are likely capable of assembling huge poxviruses such as variola, the causative agent of smallpox.
While it’s unlikely that any of the people capable of acquiring infectious samples will decide to unleash a pandemic in a given year, history strongly suggests that someone eventually will. In the 50 years since the dawn of recombinant DNA, at least one murderer possessed a background that today would confer the necessary skills: the cult member and virologist Seichi Endo, who sought to obtain the Ebola virus and later helped commit mass murder against civilians using chemical weapons. His existence alone suggests that we should expect to see one deliberate pandemic event every 50 years. Other terrorists and mass murderers who may have received a relevant education or could plausibly have acquired the necessary training include multiple al Qaeda operatives, a neuroscience graduate student who opened fire in a crowded theater, and the Unabomber, a brilliant Berkeley mathematics professor who wrote of the “immense power of biotechnology” and sought to bring down industrial civilization.
What these historical murderers lacked, and we still do not yet know, is which specific viruses are likely to cause new pandemics. The genomes for virtually all known human pathogens are freely available online, but the variola smallpox, which causes smallpox, is the sole disease that almost certainly would spread if released. Fortunately, it is also one of the least accessible viruses, and one that societies have successfully controlled and eradicated using already-stockpiled vaccines. Other frequently-cited agents include the 1918 influenza virus, MERS-CoV, and Nipah virus, but any terrorist who attempts to cause global catastrophe by unleashing one of these is likely to fail. MERS-CoV and Nipah have spilled over on numerous occasions and failed to spread indefinitely, suggesting that they are not currently pandemic-capable. Viruses such as 1918 H1N1 influenza would face preexisting population-level immunity because most people have recently been exposed to related strains that share at least one surface marker. But as soon as anyone credibly identifies novel viruses that are likely capable of igniting new pandemics and adds them to a public list of the greatest threats—the goal of many well-meaning nonprofits and government agencies—the number of individuals capable of single-handedly causing more than a million deaths will grow dramatically.
For maximum effect, a nefarious person or group probably wouldn’t release a pathogen in just one place. The best public health systems in the world—even with draconian lockdowns—couldn’t reliably stop a pandemic virus released in multiple travel hubs. Though COVID vaccines were produced in record time, even the 100 Days Mission vaccine moonshot wouldn’t match the speed of a future omicron, which spread from a single region in Africa to infect a quarter of Americans within 100 days of being identified. Since a pathogen released in multiple travel hubs would spread much faster, even rapidly produced biomedical countermeasures would arrive too slowly. And if a virus is sufficiently contagious and lethal, workers with inadequate protection may reasonably decline to risk their lives, causing food and power distribution networks to collapse—and with them, societies.
The roadmap to defense. On the surface, our prospects for indefinitely avoiding a deliberate release scenario appear bleak. But there is a silver lining. Public officials have long taken national defense issues far more seriously than public health, and our current vulnerability to malicious biology is an eminently solvable problem when paired with the resources of an organization like the Pentagon.
The first step is to delay pandemic proliferation for long enough to build adequate defenses. That starts with avoiding the handful of laboratory experiments that can assess pandemic capability, none of which is required to develop countermeasures. Unfortunately, this is not the direction that well-meaning US health and development agencies are moving in. The US National Institutes of Health (NIH) funds controversial efforts to learn which mutations could render highly lethal but poorly transmitted zoonotic viruses capable of causing a pandemic, while the US Agency for International Development (USAID)—which deserves tremendous credit for recognizing the importance of pandemic prevention long before COVID-19—aims to learn which animal viruses that haven’t yet spilled over could cause a pandemic if they did. Its DEEP VZN program doesn’t just aim to discover and sequence new animal viruses, which is a comparatively low-risk endeavor that can help develop broad-spectrum vaccines and antivirals, but also to characterize them in the laboratory to assess pandemic capability and add the winners to a public list, rank-ordered by perceived threat level. Because these agencies require their researchers to share complete viral genome sequences, any successful pandemic virus identification effort will lead to widespread access.
But health and development agencies are not the only ones interested. The international community has successfully kept nuclear weapons from falling into the hands of terrorists for 77 years. Once fully cognizant of how many people could obtain infectious samples of any identified pandemic viruses, security agencies, state departments, and legislators may step in. For example, defense agencies could request external security reviews of pandemic-related research, while state departments could push for a new pandemic test-ban treaty that would narrowly prohibit the handful of experiments capable of substantially increasing our confidence that any given virus will cause a pandemic.
Congress could address catastrophic risks by clarifying that individuals and institutions will be held liable for direct and indirect actions leading to any catastrophe causing more than a million American deaths—including sharing blueprints later used by terrorists—and requiring insurance to cover at least part of this liability. There is precedent: While capped by the Price-Anderson Act, nuclear power plant operators are already liable for the consequences of terrorist sabotage. Catastrophe liability and insurance can ensure that low-probability, high-consequence risks are factored into decision-making, allowing formal insurer risk assessments to shift costs from taxpayers to those who would perform or fund the research. Any project posing so much catastrophic risk that funders are unwilling to cover the extra insurance premiums should not proceed.
Another way to delay a deliberate release scenario is to ensure that only legitimate researchers can access synthetic DNA permitting catastrophic misuse. Hundreds of companies sell synthetic genes, reagents that increasingly power the bioeconomy, but many smaller providers do not screen orders for hazardous sequences that could be readily misused. This is primarily due to the substantial cost of hiring expert humans to scrutinize the false alarms generated by current similarity search algorithms. Newly precise exact match screening methods that choose random critical fragments of hazardous sequences, predict functional variants, and remove any that match harmless genomes can fully automate screening while making it extremely difficult for adversaries to evade. Crucially, these methods employ oblivious cryptography, which screens orders without disclosing what is ordered or considered hazardous, in order to protect trade secrets and control information that could be used by terrorists. Secure screening can also be built into next-generation “benchtop” synthesis and assembly machines that many expect future laboratories will use to make their own DNA, thereby ensuring that every synthesis company and local device will refuse to produce the genomes of new pandemic agents without authorization. The Secure DNA Project intends to make screening freely available in 2023, which may reduce unauthorized access by up to a hundredfold.
Eventually, these delaying tactics will fail. When they do, reliable detection and defensive systems must be capable of fending off the worst-case scenarios—especially those that could otherwise cause civilizational collapse. The first collapse scenario has already been described: a highly lethal and contagious pandemic could disrupt essential services, leading to inadequate food, water, power, and law enforcement.
The second scenario is more subtle, and arises from viruses that can spread without causing any clinical symptoms. Consider HIV, which can remain contagious for years before causing rapid decline and death in untreated individuals. A faster-spreading pathogen that causes a disease characterized by a long asymptomatic period could spread quickly enough to infect most of the world before anyone starts showing symptoms. Without a swiftly developed cure, nations may not be able to recover.
The solution to subtle biological threats is a detection system tuned to recognize the universal signature of all pandemic agents: their ability to spread exponentially. Untargeted metagenomic sequencing can flag all such exponentially growing sequence fragments for close analysis. Once a pathogen is identified as harmful, the extent of its spread can be monitored using targeted wastewater sequencing and rapid nucleic acid diagnostics. Ideally, defense agencies, public health systems, and philanthropies will all build and operate such detection systems in order to provide the world with as much early warning as possible.
Once a threat has been recognized, the challenge is to halt infections, especially in the essential workers who would need to continue to distribute food, water, power, law enforcement, and health care during a catastrophic pandemic. They are likely to judge N95s, which offer at best 95 percent protection, to be inadequate in a high-lethality scenario. That makes defense against the overt threat straightforward: Create comfortable personal protective equipment (PPE), demonstrate that it reliably prevents untrained users from being infected with the most contagious pathogens known without requiring fit-testing, and arrange for delivery to all essential workers within days of a new pandemic, thereby preventing the worst-case scenario. For example, a powered respirator protects all mucus membranes from exposure by combining a transparent helmet with an air pump that delivers sterile air. Crucially, it can be worn by untrained users without fit-testing. Even at $250 per unit, protecting 40 million essential workers in the United States would cost $10 billion—a pittance when compared to investments in nuclear security, especially since this intervention would directly solve the problem.
But perhaps the most promising defense is passive.
Wavelengths of light below 230 nanometers are strongly absorbed by proteins and do not appear to penetrate the surfaces of human skin and eyes. That makes “low-wave” or “far-UVC” light hundreds of times safer than slightly higher wavelengths, yet still germicidal to single-celled pathogens and viruses. At the current levels approved by the American Conference of Governmental Industrial Hygienists, overhead fixtures emitting 222 nanometer light can eliminate 90 percent of airborne pathogens each minute – significantly better than aircraft ventilation systems that filter one volume of air every three minutes – while also cleansing surfaces. And preliminary studies suggest that it might even be safe at levels that would inactivate 90 percent of viruses within one second, which is fast enough to inactivate airborne viruses emitted from one person’s mouth before they spread to a conversation partner. If safety and efficacy can be confirmed to the world’s satisfaction—and especially if we can develop efficient and inexpensive low-wave LED lights—ubiquitous installation wouldn’t just block transmission of measles, the most contagious known pathogen. It would locally eliminate virtually every other respiratory and contact-transmitted infectious disease, from the common cold to influenza. The resulting productivity gains could pay for installation many times over, suggesting a market-based path to improved public health.
All of these threat-agnostic detection and defensive systems are clearly within the purview of defense agencies, which traditionally excel in physical technologies and engineering. In contrast, the traditional vaccines and antivirals relied upon by health agencies would take too long to produce and distribute in the event of a deliberate high-lethality pandemic. Drawing on lessons from cybersecurity, we can also expect biomedical countermeasures to be less reliable because they allow adversarial agents to infect critical systems—in this case the human body. Still, when the threat is catastrophic, it’s wise to invest in layered defenses, especially when we are uncertain of the exact form of future attacks, which is why vaccines will still be a vital part of our defensive portfolio, especially if broad-spectrum or rapidly manufactured on-site. It’s encouraging that the recently released US National Biodefense Strategy doesn’t just articulate the need for faster vaccines, but also includes many elements of the roadmap that I developed with the Geneva Centre for Security Policy. The world needs them all, and haste is paramount.
Blueprints for newly identified pandemic agents could be posted at any time. The pace of recent advances in biotechnology, including self-spreading constructs such as CRISPR-based gene drive systems, suggests that it would be foolish to bet against the development of new and unanticipated pandemic-class agents. The task for society is to assume that pandemic proliferation is coming, to delay its onset for as long as possible, to build systems capable of detecting all catastrophic threats, and to prepare comprehensive defenses against deliberate attacks by competent adversaries.
The Bulletin elevates expert voices above the noise. But as an independent, nonprofit media organization, our operations depend on the support of readers like you. Help us continue to deliver quality journalism that holds leaders accountable. Your support of our work at any level is important. In return, we promise our coverage will be understandable, influential, vigilant, solution-oriented, and fair-minded. Together we can make a difference.
Other measures that would avoid a deliberate release scenario is to require virologists to determine the subspecies of the bats themselves, before allowing those bats coronavirus genomes to be filed with NCBI and GISAID. The coronavirus sequences filed with NCBI are less likely to be ‘enhanced’, if the bat subspecies for the particular collected sample is authenticated. Bat subspecies are determined by the analysis of cyt B genes extracted from bat samples. “The complete mitochondrial cyt b gene (1,140 bp) was amplified and sequenced. These sequences have been submitted to GenBank and accession numbers are shown in the Table…. For… Read more »