Two case studies in using export controls to prevent bioterrorism
Introduction
The use of advanced bioweapons is increasingly recognized by the United States government as one of the most serious threats caused by the development of emerging technologies [1]. There are two types of threat actors we may worry about building bioweapons:
- Lower-resource actors that use mainly low-skilled, low-cost methods to produce known pathogens.
- Well-resourced actors designing novel pathogens in cutting-edge research labs.
We consider two case studies in how export controls might be able to slow down each of these actors, respectively, by making it harder to obtain benchtop DNA synthesizers and cryogenic electron microscopes (cryoEM).
Benchtop DNA synthesizers are emerging near-consumer-grade technology that allows actors to synthesize pathogen DNA without any oversight, and cryoEM is an important part of the cutting edge build-test-learn loop to develop new pathogens. This report considers each of these technologies in turn.
- We first describe what the technology is in more detail before discussing the current export control regulations.
- We then provide, for each technology, an assessment of whether it is more likely to increase biorisk from high-resource or low-resource actors.
- This then informs recommended updates to export controls that minimize disruption to worthwhile research and economic activity while preventing the proliferation of dangerous bioweapons.
Benchtop DNA synthesizers
Modern benchtop DNA synthesizers could enable faster, easier development of dangerous pathogens. They could also allow researchers to manufacture synthetic genes privately, evading current safety screening protocols that are implemented via centralized DNA synthesis providers.
Rapid DNA synthesis is already within easy reach of a well-resourced actor, so widespread availability of benchtop synthesizers is more likely to democratize access to genetic engineering than it is to significantly improve the capabilities of current biotechnology leaders. Consequently, export controls on benchtop synthesizers should focus on the prevention of bioterrorism by rogue states and extremist groups, not on hobbling the bioeconomies or bioweapons capabilities of rival nation-states.
Benchtop synthesizer technology
DNA synthesis is the ability to manufacture custom genetic instructions to change the behavior of viruses and living organisms. It is crucial to modern biotechnology research and production and has enabled breakthroughs in medicine, clean manufacturing, and agriculture.
Currently, most biotech researchers order strands of synthetic DNA from centralized providers and then assemble them in the lab into longer stretches of DNA that encode synthetic genes. Although benchtop synthesizers, which can manufacture DNA on-demand in a lab, are available, they are costly, and their use is not widespread.
However, recent innovations may rapidly increase the popularity of benchtop synthesizers in coming years.
- New chemical methods of DNA synthesis are making benchtop synthesizers easier to use, as well as enabling them to manufacture longer stretches of DNA more accurately.
- In the next 1-2 years, multiple companies plan to release benchtop DNA synthesizers that also perform automated DNA assembly[2], which will further automate the process of creating synthetic genes and likely make benchtop synthesizers more appealing.
- Costs of benchtop synthesizers are expected to fall as companies compete to develop new models.
Biosecurity implications of benchtop synthesizers
Access to benchtop synthesizers can reduce a time-consuming step in the build-test-learn cycle of biotech research, and it could also provide scientists with the ability to keep their synthetic DNA sequences confidential from DNA synthesis companies. For well-intentioned researchers, this could increase the speed of productive biotech breakthroughs and ensure the security of valuable intellectual property.
However, it is possible that the new generation of benchtop DNA synthesizers+assemblers could significantly decrease the degree of expertise required to manufacture the genome of a dangerous pathogen. While there are still significant technical hurdles to overcome to turn synthetic genomes into a weaponized version of the pathogen[3], automated gene synthesis would eliminate an important barrier, potentially increasing the number of individuals capable of creating a powerful bioweapon.
In addition, most centralized DNA synthesis providers have screening systems that prevent customers from ordering dangerous genes without sufficient proof of legitimate research needs. The recent Executive Order on Safe, Secure, and Trustworthy AI enshrines this practice by making such screening mandatory for any researchers receiving federal funding[4]. However, widespread availability of benchtop synthesizers could provide a way for researchers to evade this screening, removing an important safety check against nefarious behavior.
Current export controls will extend to next-generation benchtop synthesizers
US export policy currently imposes some restrictions applicable to benchtop DNA synthesizers. US biosecurity-related export policy is based on the recommendations of the Australia Group, a group of countries that agree on trade regulations to prevent trafficking and development of chemical and biological weapons, of which the US is a member. The Export Administration Regulations (EAR) and Commerce Control List (CCL) currently require a license for export of certain DNA synthesizers and assemblers. They also require licenses for the export of any software designed to be used with these synthesizers/assemblers to produce DNA sequences from digital genetic data[5].
Requirements for a DNA synthesizer or assembler to fall under these restrictions[6]:
- Partially or entirely automated
- Designed to generate continuous nucleic acid sequences greater than 1.5 kilobases in length with error rate less than 5% in a single run.
Current benchtop synthesizers are not yet able to produce 1.5 kilobases with a <5% error rate, but it is expected that the next generation of products will fit these criteria.
The export restrictions on DNA synthesizers fall under both the “CB” (Prevention of chemical and biological weapons development) and “AT” (Anti-terrorism) justifications in the EAR system. The CB restricted country list prohibits unlicensed export to all countries not in the Australia group, whereas the AT list is much shorter, including only countries the US considers state sponsors of terrorism, including Syria, North Korea, and Iran.
In addition, US persons are forbidden from exporting or transferring these DNA synthesizers or accompanying software without a license if they know these tools will be used in the design, production, or stockpiling of biological weapons in any country.
Will current export controls meet the biosecurity challenges posed by benchtop synthesizers?
When benchtop DNA synthesizers become capable of synthesizing 1.5 kilobase stretches of DNA, the above export regulations will apply, and United States manufacturers will need a license to export benchtop synthesizers outside of the Australia Group countries. The most advanced benchtop synthesizers are currently made by companies in the United States and by close allies who also participate in the Australia Group[7], so depending on the dissemination speed of this technology, enforcement of this restriction could limit non-Australia Group nations’ access to next-generation benchtop synthesizers for several years.
Impact on well-resourced, state-sponsored bioweapons programs:
Rapid DNA synthesis is already accessible to any well-resourced biotech lab, either by ordering from a DNA synthesis provider or by using an older benchtop synthesizer and assembling the shorter DNA fragments in the lab. For example, if we imagine a state-sponsored Chinese lab working on cutting-edge research, they would be a high-priority customer for a Chinese DNA synthesis provider (of which there are many), and they would have plenty of scientists and lab robots to enable quick assembly of the resulting synthetic DNA.
Consequently, restricting access to benchtop synthesizers will probably not substantially reduce risk from powerful rival states’ bioweapons programs. Restrictions might inhibit the progress of bioweapons programs run by smaller, less wealthy nations, but not the progress of a well-resourced adversary with its own extensive biotech sector.
Impact on lower-resourced groups pursuing bioterror agendas:
Inexpensive next-generation benchtop synthesizers could substantially decrease the barrier to bioweapons development for lower-resource states or terrorist groups. The question, then, is whether the existing export controls would be effective at preventing these groups from accessing benchtop synthesizers.
In the case of low-resourced states for which there is strong evidence of nationally-sponsored bioweapons programs, such as North Korea, a blanket export ban seems appropriate. However, for most countries, the national governments themselves are not the primary source of risk. Rather, the danger comes from non-state terrorists either located within, or using supply chains running through, those nations. The risk of exporting benchtop synthesizers to one of these nations is determined by how well that nation prevents terrorist groups from buying and using scientific equipment within its borders.
Export of benchtop synthesizers to known terrorist groups and military end-users are restricted under current regulations, but we know of no easy way to determine how good individual countries are at spotting questionable use of biotech equipment within their borders. Measures such as required DNA synthesis screening seem applicable, as well as local law enforcement and counter-terrorism activities (which themselves can be a double-edged sword.) We hope to assess this further in future work.
Proposed changes to export controls
Since the most pertinent risks of benchtop DNA synthesizers relate to enabling low-resource, rather than high-resource, bioweapons programs, we believe that some of the current nation-wide export controls to high-resource nations may be too stringent. The restrictions’ risk reduction in these cases may not be worth disruption to legitimate economic activity and scientific progress.
On the other hand, we believe that the existing fine-grained controls designed to prevent smaller-scale terrorist action are likely insufficient. While they may prevent benchtop synthesizers from being exported to known terror groups, smaller terrorist operations won’t likely be known internationally. In addition, distributed, agile terrorist networks will likely find it possible to circumvent regulations.
It is diplomatically important to reward Australia Group countries for their participation by granting them greater freedom to trade with the United States in biotech materials and equipment, so no additional restrictions should be either lifted or added without consulting them. To address the above issues, we believe the United States should propose to the Australia group that:
- Internal restrictions on remote sequence screening for benchtop synthesizers be implemented in each Australia Group country.
- Australia Group countries should draft a know-your-customer checklist for businesses that want to export cutting-edge biotech equipment.
- Once these two measures are implemented, the Australia Group as a whole should relax restrictions on DNA synthesizer export to other nations, excluding only nations with known bioweapons activity or sponsorship of terrorism.
By implementing these steps, we hope that benchtop synthesizers with built-in safety measures will become the most cost-effective and appealing option throughout the world.
In addition, we hope that these protocols will provide a broadly-applicable template for defending against small-scale, low-resource bioweapons as the enabling technology becomes cheaper and more accessible.
Cryogenic Electron Microscopes
CryoEM: A cutting-edge technology for developing biomolecules
“Cryogenic Electron Microscopy” (cryoEM) is a tool used to study the 3D structure of biomolecules such as pandemic agents. Its name derives from the two technologies it uses:
- Electron Microscopy: A technique used to take extremely high-resolution microscope images of samples. A traditional microscope illuminates a sample with light, and then uses lenses to magnify the image viewed by your eyeball (or a camera). To a rough approximation, an electron microscope instead “illuminates” the sample with electrons, and uses magnetic fields to make “electron lenses” that magnify the image viewed by a camera.
- Cryogenics: For biological samples, it can be difficult to keep molecules stationary while being imagined as well as keep them cool while being irradiated with electrons
3D structure data can be determined from cryoEM data by using sophisticated image analysis software. The cryoEM takes several (e.g. >10,000) images which display different perspectives on the molecule (edge-on, face-on, flipped upside-down, etc.), as well as several ways the molecule could be distorted. This is a computationally difficult challenge to then back out what the full 3D structure must look like that can produce all these perspectives and distortions. These machines are very specialized and complicated, and their invention was awarded the 2017 Nobel Prize in Chemistry. A photograph of one is shown in Figure 1. These machines cost on the order of millions of dollars, have a several month lead time, and require thousands of dollars of operational costs per day (though there are efforts to reduce these costs, see Hand 2020). Therefore, they are usually only purchased by well-resourced actors that are doing cutting-edge research.
While alternative technologies to cryoEM do exist (such as “x-ray crystallography”), its importance to developing biomolecules can be seen from, e.g. its use by leading drug development companies[8]. This example also highlights the inherently dual-use nature of biotechnology, cryoEM could help facilitate the development of a live-saving drug or a deadly pathogen[9]. Therefore, it merits further investigation to understand how to control the use of cryoEM, but at this time it is not understood the extent to which this would slow malicious actors or the broader effects this would have on the US economy or beneficial uses of biotechnology. The rest of this piece will explore how these export controls could be crafted and evaluated.
CryoEM is (probably) not export controlled
There are several relevant authorities that determine export controls that might affect cryoEM, including:
- United States Commerce Control List, especially the following categories:
- United States Munitions List, especially Category XIV
- Australia Group (multilateral export controls focused on chemical and biological weapons)
- Wassenaar Arrangement (multilateral export controls on items including dual-use goods and technologies)
Of these, the only list that includes electron microscopes at all is the United States Commerce Control List Category 3, and these are specifically microscopes used for inspecting semiconductor equipment and explicitly exclude general purpose scanning electron microscopes (e.g. 3B991.b.1.j, “Electron beam systems “specially designed” or modified for mask making or semiconductor device processing”).
Therefore, we do not think that cryoEM is currently export controlled (though note that I am very much not a lawyer).
CryoEM is not necessary to build a bioweapon, but, if we did want to export control it, there are some potential “chokepoints”
If the US government decided that it wants to export control cryoEM, there are several challenges and opportunities that need to be understood before putting this into regulation.
Challenges
One class of challenges to export controls surrounds the substitutability of the technology. As briefly mentioned above, cryoEM is not the only technique that can be used to understand the 3D structure of a biomolecule, other technologies like x-ray crystallography and nuclear magnetic resonance (NMR) could also provide these insights (though they also bring their own challenges). Even more broadly, it is not clear the extent to which measurements with any technique will be necessary to determine the 3D structure of a molecule as computational structure prediction (such as Google DeepMind’s AlphaFold) continues to improve. Finally, even if the above challenges could be overcome, this would still not prevent malicious actors from building bioweapons that use pathogens with known 3D structures (e.g. minor mutations on a known pathogen), or that do not require any research and development at all (e.g. an existing pathogen).
Opportunities
Given that cryoEM is relatively specifically used for 3D structure of biomolecules, and it involves relatively sophisticated hardware and software, it is plausible that export controls could be crafted in a way that minimally affects tools that are trailing edge or not used for biotechnology at all (such as conventional electron microscopy). Understanding the supply chain well enough to make these determinations are beyond the scope of this work, but two potential places to look for “chokepoints” controlled completely by US or allied countries are:
- The “direct electron detector device (DDD) cameras” that revolutionized the spatial resolution achievable with cryoEM
- The software used to analyze the collected images
Recommendations
- Do not export control cryoEM at the present moment
- Do more research in to the supply chain to identify “chokepoints” and which nations control them
- Should intelligence indicate that a well-resourced actor might have a bioweapons programs, restrict export of cryoEM to them
Footnotes
For example, increased attention to the threat of bioweapons is evident in the 2018 Export Control Reform Act, the 2023 creation of the Office of Pandemic Preparedness and Response Policy, and the 2023 Executive Order on Safe, Secure, and Responsible AI.
US-based Telesis Bio has announced a 2024 release for a combination benchtop DNA synthesizer and assembler, and UK-based Evonetix and Nuclera are also developing benchtop synthesizer+assemblers.
Benchtop DNA Synthesis Devices: Capabilities, Biosecurity, and Governance. NTI, 2023. https://www.nti.org/wp-content/uploads/2023/05/NTIBIO_Benchtop-DNA-Report_FINAL.pdf
Section 4.4 Executive Order on the Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence. United States Executive Office of the President, 2023. https://www.whitehouse.gov/briefing-room/presidential-actions/2023/10/30/executive-order-on-the-safe-secure-and-trustworthy-development-and-use-of-artificial-intelligence/
Commerce Control List, 2D352.
Export Administration Regulations, 2B352.j.
Notable companies currently selling or developing next-generation benchtop synthesizers: Telesis Bio (US); DNA Script (France); Evonetix (UK); Nuclera (UK).
Spot-checking four drug design companies, all of them currently work with cryoEM: Roche, Novartis, Merck, Pfizer.
For a more explicit example, one company, Generate Biosciences, is using cryoEM to make a training set for a generative AI model that can make, among other things, “stealth proteins” (that are a class of agents that may be able to cure cancer, but also seem like they could plausibly make more deadly diseases)