Introduction

Medical diagnostics, biomanufacturing, and many other parts of the bioeconomy rely on synthetic DNA and RNA purchases, which are ordered from commercial providers and shipped to laboratories around the globe. In addition to enabling beneficial biotechnology, affordable and accessible nucleic acid synthesis raises biosecurity concerns: some sequences can be used to reconstruct pathogen genomes or engineer dangerous biological agents, and it’s necessary to ensure those sequences are not misused by actors seeking to cause harm.

Most commercial synthesis providers screen the orders they receive to identify sequences of concern that could facilitate the construction of dangerous biological agents. When sequences in an order are flagged, follow-up screening determines whether the order is fulfilled. This screening centers on the customer: do they have a legitimate, peaceful purpose for obtaining the flagged sequences of concern?

This follow-up screening is the subject of this briefer. Between July and August 2023, we interviewed industry contacts and other policy and biosecurity experts. In the subsequent months, we conducted independent research and solicited expert feedback on report drafts. This process made clear that today, the follow-up screening process is ad-hoc. The customer service representatives, bioinformaticians, and security experts conducting follow-up screening often lack support for handling ambiguous cases, and there is little infrastructure to support information-sharing with other synthesis providers or law enforcement. This is a serious security concern. As such, we recommend that the US government take the following steps:

1. Support the creation of third-party services for Know Your Customer (KYC) screening, including developing standards for customer legitimacy
2. Support the creation of software tools to assist decision-making during follow-up screening of flagged orders by offering context such as red flags, customer data, and order history
3. Standardize data formats for records of flagged orders, contributing to improvements in screening algorithms, decision support tools, and information-sharing
4. Require information about flagged orders to be securely shared with a network of other providers and law enforcement
5. Require red-flagged orders to have a second approver from the same institution, such as a biosafety officer, mitigating some accident risks and insider threats

This briefer begins with an overview of current approaches to follow-up screening and the gaps and challenges associated with these approaches. We then detail recommended actions the US government can take to support comprehensive and robust follow-up screening.

---

Current Follow-Up Screening Approaches

Follow-up screening of flagged orders is a critical and complex part of securing nucleic acid synthesis. It is not always straightforward to distinguish customers with a legitimate, peaceful purpose for sequences of concern from customers who would misuse the sequences.

All nucleic acid synthesis orders should undergo both sequence and customer screening, according to current best practices, such as the US Screening Framework Guidance for Providers and Users of Synthetic Nucleic Acids¹ and the International Gene Synthesis Consortium (IGSC) Harmonized Screening Protocol.²

For some orders, sequence or customer screening raises flags that must be resolved by follow-up screening. Sequence screening aims to flag sequences of concern (SOCs) that contribute to toxicity (e.g., botulinum toxins), pathogenicity (e.g., proteins that can neutralize antibodies), or that are from regulated biological agents (e.g., the 82 pathogens that appear on the Australia Group Common Control list.³) Customer screening may raise red flags, such as names appearing on blocked persons lists, requests for anonymous payment, or unusual shipping procedures.⁴

Follow-up screening determines whether the flagged order is fulfilled based on verifying the legitimacy of the customer and their order. The US Screening Framework Guidance recommends that any customer ordering a sequence of concern be verified as “a legitimate member of the scientific community.” The IGSC protocol implements this by requiring customers to “provide a written description of the intended use of the synthetic product,”⁵ after which providers “verify independently a) the identity of the potential customer and purchasing

organization, and b) that the described use is consistent with the activities of the purchasing organization.”⁶

If follow-up screening does not resolve concerns, providers are encouraged not to fulfill the order and to report it to law enforcement, specifically their local FBI field office’s weapons of mass destruction coordinator. The IGSC protocol states that members should have “established relationships with local and national law enforcement and intelligence authorities” to whom they can report potentially problematic orders. This process is shown in Figure 1 below.

Figure 1: Workflow for screening of nucleic acid synthesis orders. This figure reflects best practices in the US Screening Framework Guidance and IGSC Harmonized Screening Protocol. If either sequence or customer screening flags an order, follow-up screening seeks to verify legitimacy. Orders should only be fulfilled after follow-up screening resolves any concerns. If concerns remain, information should be shared with law enforcement.

Follow-up screening, per a March 2023 paper, “is the most time-consuming step in DNA synthesis screening. It typically requires a staff member with a PhD in bioinformatics or similar training to perform the work.”⁷ It is not a rare occurrence; an April 2022 policy paper from the Engineering Biology Research Council (EBRC) notes that “an estimated 5% of orders are flagged for review. ‘Yellow’ flag hits might take one to two hours to resolve, and ‘red’ flag hits can take several hours.”⁸ Given the continued growth of the synthesis industry and the recent expansion of the Screening Framework Guidance to cover shorter sequences and additional transactions, as well as its critical role in securing nucleic acid synthesis, supporting follow-up screening should be a priority for the US government.

As this report was being prepared for publication, the White House issued an “Executive Order on the Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence.”⁹ This Executive Order tasked the Office of Science and Technology Policy (OSTP) with establishing a framework “to encourage providers of synthetic nucleic acid sequences to implement comprehensive, scalable, and verifiable synthetic nucleic acid procurement screening mechanisms,” including “customer screening approaches to support due diligence” and “processes for the reporting of concerning activity to enforcement entities.”¹⁰ Both customer due diligence and reporting concerning activity occur during the follow-up screening process, and the mechanisms for screening described in this report could be included in the framework established by OSTP.

---

Gaps in Follow-Up Screening

There are many gaps in the current process for follow-up screening. Industry contacts report that the screening process is conducted in an ad-hoc fashion without specialized tools (e.g., looking up a customer’s name on LinkedIn to confirm that they work at the lab listed in their order), that it is unclear how to verify legitimacy in many ambiguous cases (e.g., verifying whether an industrial customer with no academic history has a legitimate need for a sequence of concern), and that there is little infrastructure to support information-sharing with other synthesis providers or with law enforcement.

Providers Have Few Tools for Follow-Up Screening

There are too few tools available to assist with verifying the legitimacy of a customer and their proposed end-use for sequences of concern.¹¹

Although a number of software tools are available to flag sequences of concern, these tools do not offer sufficient context or confidence to make follow-up screening decisions about flagged sequences. Even if a flag is contextualized with the organism, a sequence appears to be sourced from, and guidance on the level of concern it raises (i.e., a “red” vs. “yellow” flag), industry contacts report that bioinformatics experts must manually assess sequences of concern¹². For example, deciding whether flagged sequences are consistent with the proposed use described by the customer relies on heuristics built up over years of experience rather than codified standards. Software tools to support these assessments would allow experts to make screening decisions more quickly and consistently and, in some cases, without a bioinformatics expert in the loop.

Software developed to support “Know Your Customer” processes outside the synthesis industry can assist synthesis providers with some steps of customer screening, such as verifying customer identity documents and checking customers’ names against government watch lists. However, industry contacts report that verifying even basic information about a customer’s academic or business affiliation is not straightforward, given the global variety of company and laboratory documentation and the difficulty created by language barriers.¹³ During follow-up screening, providers typically glean information from their internal Customer Relationship Management (CRM) software, designed to track client relationships rather than supporting screening, from searching the internet for public information and contacting customers directly.

The reliance on ad-hoc tools that are not fit for purpose means the providers expend more time and effort for less reliable conclusions. A recent report from the Nuclear Threat Initiative notes that customer screening by a third party might alleviate some challenges but that “currently there is no consensus on what constitutes a legitimate user, and there are no entities that offer such a service.”¹⁴

No Verifiable Standards for Customer Legitimacy

There is no established consensus for what constitutes a legitimate customer for synthetic nucleic acids, making it challenging to design a follow-up screening process and verify whether screening is effective. Longstanding debates about dual-use in biology have highlighted the difficulty of distinguishing efforts to develop biological weapons from some legitimate and peaceful research, such as work with medical toxins,¹⁵ viral engineering,¹⁶ and synthetic biology.¹⁷

The Administration for Strategic Preparedness and Response (ASPR) within the Department of Health and Human Services (HHS) issued a Companion Guide to Assist in Implementing the Recommendations of the Screening Framework Guidance in 2023.¹⁸ It includes some illustrative use cases and red flags for verifying legitimacy but stops short of defining who is considered legitimate. The listed red flags include easily recognizable indicators of suspicious behavior, such as requests to arrange payment through a third party or to misidentify goods on packaging. Other scenarios would require more detailed investigation, such as verifying that the university a customer claims to be affiliated with has records of their employment.

A few examples of ambiguities for verifying legitimacy that a standard could resolve are:

• Is a LinkedIn profile or personal website sufficient evidence to support a customer’s claims about their affiliation, given that this could be easily spoofed?
• Is a customer required to have a virology research history to be provided with a viral sequence of concern, or is it sufficient to verify their affiliation with a life sciences institution?
• Does a customer need to provide evidence that their startup works on diagnostics related to a sequence of concern, or is it sufficient to verify that the startup is a registered company?
• Are non-traditional laboratory spaces, like high school science labs or community biology labs, acceptable places to ship sequences of concern?

A well-defined standard for customer legitimacy would be less error-prone, less time-consuming to verify, and easier to operationalize across the nucleic acid synthesis industry. This kind of verifiable standard would reduce the burden on providers by creating greater clarity in cases that are currently ambiguous. Standards for screening would also make it more difficult for customers to subvert security by bringing orders rejected by one synthesis provider to another with less stringent screening.

Little Infrastructure to Support Information-Sharing

Some patterns of suspicious behavior can only be detected if providers are able to share order information with one another. For example, customers may seek to evade sequence screening by splitting orders between multiple providers or switching from one provider to another in response to follow-up questioning. The IGSC has an information-sharing mechanism intended to assist providers in recognizing suspicious customers, but it has rarely been used as of this writing. An EBRC policy paper states that it’s not known if this is because concerning orders are rare or because businesses are not comfortable sharing sensitive information with potential competitors.¹⁹

Information-sharing infrastructure could also support reporting to law enforcement entities. When follow-up screening does not resolve malicious intent concerns, the Screening Framework Guidance and IGSC Harmonized Screening Protocol recommend reporting the order to law enforcement. This reporting involves informal phone calls between providers and their local FBI field office’s weapons of mass destruction coordinator. Better information-sharing infrastructure would allow enforcement entities to receive more reports and better analyze the content of reports.

---

Recommendations for Supporting Follow-Up Screening

The US government is well-placed to fill these gaps in follow-up screening. We recommend five actions for the US government to take to support follow-up screening of flagged nucleic acid synthesis orders:²⁰

1. Support the creation of third-party services for Know Your Customer (KYC) screening, including developing standards for customer legitimacy.
2. Support the creation of software tools to assist decision-making during follow-up screening of flagged orders by offering context such as red flags, customer data, and order history.
3. Standardize data formats for records of flagged orders, contributing to improvements in screening algorithms, decision support tools, and information-sharing
4. Require information about flagged orders to be securely shared with a network of other providers and law enforcement.
5. Require red-flagged orders to have a second approver from the same institution, such as a biosafety officer, mitigating some accident risks and insider threats.

1. Support the creation of third-party services for Know Your Customer (KYC) screening, including developing standards for customer legitimacy

The US government is well-positioned to create or support the creation of a third-party service to perform Know Your Customer (KYC) screening.

Standard methods for determining customer legitimacy will be needed, and developing these standards is a key way for the government to support KYC screening. Standards for customer legitimacy could be developed through expert studies conducted by the National Academies of Sciences, Engineering, and Medicine (NASEM), by the National Institute of Standards and Technology (NIST), or through an independent blue-ribbon commission.

The customer screening steps recommended in the HHS Screening Framework Guidance and IGSC Harmonized Screening Protocol could be used as the basis for an initial definition. The Harmonized Screening Protocol states that IGSC members “require identification data from all potential customers,” as well as “supply genes from regulated pathogens only to researchers in bona fide government laboratories, universities, non-profit research institutions, or industrial laboratories demonstrably engaged in legitimate research,” and“verify independently…that the described use is consistent with the activities of the purchasing organization.”²¹

Once standards for customer legitimacy have been developed, third-party KYC services could ultimately be provided by the government, similar to National Instant Criminal Background Check System (NICS) namechecks used for firearm sales or the private sector, similar to DTCC’s Avox Data Services used by the financial industry as a reference for legal entity data. Contacts noted that a government KYC service might be viewed as inappropriate surveilling of scientists but also noted that sharing customer order information with a private actor would raise IP and privacy concerns. Both governmental and non-governmental options should be explored; following the example of the financial sector, the government may be able to support the creation of a robust ecosystem of private sector KYC services through due diligence requirements and the resulting insurance incentives.

For all customers, a third-party KYC service should allow a customer’s identification to be verified and allow the customer’s name to be checked against restricted persons lists (e.g., the Specially Designated Nationals And Blocked Persons List list). Validation of supporting identification or documentation regardless of language or country of origin would be valuable; contacts described language barriers as a significant challenge for both customer and follow-up screening.

Additionally, industry contacts indicated that it would be useful for a KYC service to collate a scientific profile of customers. The current follow-up screening process often involves Googling a customer’s name to verify their publication history or institutional affiliation; this kind of public information could be collected more thoroughly and reliably by a third party and then presented to gene synthesis companies.

Scientists should not be required to undergo additional licensing to receive gene synthesis orders; several government and industry contacts we interviewed strongly recommended against this.²² However, voluntary registration with a KYC service could speed up screening and reduce repetitive demands upon customers to verify identification and provide supporting information. A voluntary KYC approach could achieve many of the benefits of synthesis screening with fewer drawbacks. The scientific community has already developed voluntary persistent digital identifiers, such as the Open Researcher and Contributor ID (ORCID), which can be associated with information relevant for follow-up screening.

As standards for legitimacy are defined, care must be taken to balance security needs against the need to encourage biotechnology innovation happening outside of traditional scientific institutions. For example, guidance must be issued on handling ambiguous cases, such as a small diagnostics startup housed in a hardware incubator space or a high school iGEM team supervised by teachers. A third-party KYC service should overall make it easier for providers to serve these non-traditional customers, as more challenging identity and safety verification only needs to be done once, the organizations performing the verification can gain more specialized expertise in this work, and the additional information required can be more standardized.

We recommend that a third-party KYC service carry liability due to lessons learned from the experience of implementing KYC screening to support anti-money laundering regulations.²³ If providers carry all the liability for information retrieved from a third party, then they will have to conduct extensive double-checks, and the service will provide limited added value. It may be challenging to convince a private or nonprofit service to accept this liability; a contact working on a nonprofit sequence screening service mentioned reluctance to provide decision support due to liability concerns. A private service maintaining such liability could require special government backing to be viable, and this should be one consideration in exploring whether it should be a government or private entity or a type of public-private partnership.

Industry contacts also emphasized the importance of providing an API for easy customer data integration with their in-house software and record-keeping. The API should be simple and well-documented and use a standard protocol (e.g., JSON or XML accessible via a RESTful architecture). If the services are not designed with integration in mind, the many providers with well-developed order management software will be unlikely to adopt it.

2. Support the creation of software tools to assist decision-making during follow-up screening of flagged orders by offering context such as red flags, customer data, and order history

Follow-up screening is the most costly part of screening, and providers currently have few tools to support this process. The burden on providers could be reduced if the government supplied or drove the creation of decision support tools to assist in triaging flagged orders and conducting follow-up customer screening. Consultations revealed the following needs for information provided by a decision support tool:²⁴

Description of why an order was flagged and a degree of concern, separating at least “red” vs. “yellow” flags.

Not all orders flagged during sequence, or customer screening represent the same degree of concern, and follow-up screening can be made less costly for less-concerning orders while allowing providers to focus their attention on riskier orders.

Rather than specifying an exact threshold of concern that should qualify as a “red flag,” we believe a flexible system should be introduced that allows the threshold to be tuned over time in response to the burdens of screening and changes in bioengineering capabilities. Standard data reporting (Recommendation 3) will enable this threshold of concern to be collaboratively tested and tuned.

For sequence screening, a starting point to express degrees of concern would include:

• Red flag: sequences of concern from potential pandemic pathogens, sequences that could endow or enhance pathogenicity to create a pandemic pathogen, orders that cover a large percentage of a regulated organism
• Yellow flag: sequences of concern from regulated organisms that do not have pandemic potential (e.g., some bacterial pathogens)
• Gray flag: sequences known not to pose biorisk concerns but that are subject to export controls or other regulatory regimes (e.g., the commonly-used 2A cleavage and IRES translation initiation sequences sourced from foot-and-mouth-disease virus)

Some existing sequence screening software already raises subjective red and yellow flags. Other sequence screening software²⁵ indicates which functions of concern (e.g., “cytotoxicity”) are present without expressing a level of concern. As the decision-support system is developed, it may become apparent that more than two degrees of concern are needed.

Customer screening can also raise different degrees of concern. For example, a request to misrepresent the contents of a package on a shipping label is more immediately concerning than a person who cannot immediately be connected to a life sciences institution. The Red Flags for Verifying Legitimacy in the Companion Guide to Assist in Implementing the Recommendations of the Screening Framework Guidance,²⁶ such as evasiveness about identity and requests to change the recipient’s name before a package is shipped, can form a starting point for separating customer screening flags into different degrees of concern.

Profile of the customer who placed the order

This profile should include publicly available data about the customer and their associated institution (LinkedIn profiles, Google Scholar or ORCID page, institution homepage, company website, etc.) and any supporting documents provided by the customer to accompany the order.

Where possible, the profile should include order history from associated customers (e.g., from the same academic lab or company) and customer information provided by a third-party KYC Service.

Order history, indicating whether this order is similar to previous orders sent to the customer

Visual representations of order history could allow analysts to quickly make judgment calls (e.g., order density across the tree of life would enable an analyst to screen a filovirus sequence to recognize if the customer has a long history of filovirus orders).

A more advanced tool would attempt to flag groups of orders that would, in combination, allow reconstruction of the genome of a dangerous pathogen (e.g., complementary sequences from the same controlled virus). This scenario is an exception to the general rule that an order is less concerning if similar orders have been approved in the past.

Sharing information between providers (Recommendation 5) could make it possible to recognize if a customer is splitting orders between multiple providers to evade detection.

An interface to capture follow-up screening data in a standard format

This interface would support record-keeping (Recommendation 3) and information-sharing (Recommendation 4). It could also support requests often made to customers during follow-up screening, such as requests for export licenses or descriptions of the proposed end use for sequences of concern.

3. Standardize data formats for records of flagged orders, contributing to improvements in screening algorithms, decision support tools, and information-sharing

Existing screening guidance recommends that providers retain records of gene synthesis screening. Standard data formats would allow records to be analyzed, contributing to improvements in screening algorithms and laying the groundwork for more automated decision support tools (Recommendation 2). Standard formats would also enable providers to move beyond an informal “one-on-one” model of information sharing (Recommendation 4).

Improvements in screening algorithms require the development of metrics that assess their efficacy, which was one of the key recommendations in an April 2022 EBRC policy paper.²⁷ Standard data formats will allow flagged orders to be systematically analyzed, revealing patterns in which orders are ultimately approved. For example, if a specific type of “red” flagged sequence of concerns is always approved with minimal follow-up screening, it may be appropriate to reduce them to a “yellow” flag. Similarly, some “yellow” flagged sequences may need to be upgraded to “red” flags based on order refusals.

What information should be stored in a standardized data format? The Screening Framework Guidance recommends keeping records of the following (the following is quoted directly from the source):

• Customer information (point-of-contact name, organization, address, email, and phone number), order sequence information (nucleotide sequences ordered, vector used), and order information (date placed and shipped, shipping address, receiver name)
• Records of protocols for sequence screening and for determining whether a sequence hit qualifies as a sequence of concern
• Records of screening documentation of all hits, even if the order was deemed acceptable
• Records of any follow-up screening, even if the order was ultimately filled
• The ultimate disposition of any SOC orders, with documentation of reasoning for the final decision (fulfill versus deny)²⁸

The IGSC Harmonized Screening Protocol also recommends keeping records of the following:

• The synthetic DNA sequence, the vector (if applicable), the recipient’s identity, and the shipping address
• Every gene sequence screening result²⁹

Based on these best practices and consultations with industry and think tank contacts about how their current processes implement these standards,³⁰ we recommend that a standard data format capture the following:

• Customer identity: point-of-contact name, organization, address, contact information
• Order metadata: date placed and shipped, shipping address, receiver name, individual placing order (often differs from customer point of contact)
• Order sequence: nucleotide sequences ordered, vector used
• Sequence screening results: protocol used, sequence(s) of concern identified, percent homology or alignments, organism(s) associated with sequences of concern
• Follow-up screening conducted: language used in back-and-forth with the customer, whether affiliation or identification was re-verified, any contact with biosafety officer, proof of appropriate import permit, confirmation of relevant research history, confirmation of proposed end-use of order
• Follow-up screening results: whether the order was filled, justification for fulfillment decision

A truly standard data format, like the XML specification of Suspicious Activity Reports under the Bank Secrecy Act, would allow information from many providers to be integrated. However, it is essential to allow some flexibility in the format, especially when it is first introduced. The IGSC’s suspicious order report infrastructure, which was designed to prevent customers from venue shopping for refused orders, has only been used once, and an industry contact speculated that this might have been because the reporting form was too detailed and time-consuming, and users were unaware they did not need to supply information in every field.³¹ To manage these issues, many fields in the data format should be clearly indicated as optional, and a field should be added for unstructured data that the provider wishes to share but is not yet included in the standard format.

4. Require information about flagged orders to be securely shared with a network of other providers and law enforcement

We recommend that when a customer or sequence screening raises a red flag about an order or whenever follow-up screening is conducted, providers send data to an information-sharing repository and share it with other providers. When follow-up screening does not resolve concerns, and a red-flagged sequence of concern is included in the order, we recommend that order data be shared with law enforcement entities (e.g., the local FBI field office’s weapons of mass destruction coordinator). If records are sufficiently standardized (Recommendation 3), providers can share data with each other and with law enforcement in a consistent and integrated way.

Integration of data from multiple providers is needed to prevent screening from being subverted by venue shopping by nefarious actors (i.e., switching orders to providers with less rigorous screening) or order splitting (i.e., splitting sequences across multiple providers to evade detection). This sort of evasion is seen in the financial industry, where money launderers channel funds through different financial institutions and across borders, as well as into channels such as methamphetamine production for which illegal purchasing rings split their over-the-counter acquisition of the meth precursor pseudoephedrine (PSE) across multiple pharmacies.

Such an information-sharing system is analogous to the National Precursor Log Exchange³² (NPLEx) system used to track PSE sales in 35 states. That system electronically logs purchases (the quantity of PSE and the customer’s name) and looks for suspicious patterns. We propose that a similar central repository for standardized data about flagged orders be established, though note that unlike PSE sales, where millions of customers make roughly uniform purchases, many synthesis orders contain valuable IP and information about which customer is ordering which DNA (and even the quantity of DNA being ordered) has significant economic value.

Security will be critical for this information-sharing repository due to concerns about intellectual property and the dual use of flagged sequences of concern. Data should be encrypted throughout the system in storage and transport (e.g., via strong public key and symmetric key cryptography, according to an existing standard for sensitive governmental information). To protect customer IP, plain language labels must be obfuscated, with access to unblinded data only available to a set of authorized government contacts or after a request for data is approved by the provider who shared it. To prevent a bad actor from using the flagged sequences in the repository as a guide for designing a dangerous biological agent, efforts should be undertaken to secure the database and to ensure that flagged sequences cannot become public through public records requests, even if the government maintains the central repository.³³

In addition to improving security, pooled data will address a core barrier to regulating the gene synthesis industry, which is a lack of standards against which providers can be certified. This recommendation, in combination with Recommendation 3, would enable the collection of datasets for which sequences are flagged, how frequently orders are flagged, and the characteristics of orders that fail to pass follow-up screening. It’s difficult to estimate the frequency at present, as gene synthesis order data is proprietary. Still, one industry contact noted that their company has only alerted law enforcement about an order a few times across a decade of operations.³⁴

Both industry and law enforcement contacts indicate that mandatory reporting makes it easier to justify data sharing to customers; providers want to tell their customers that they are compelled to share data, only do so in narrow cases, and blind details about the customer and their organization. Data should be accessible to law enforcement but should not be forwarded automatically; law enforcement officials who we contacted indicated that this could harm trust between FBI Weapons of Mass Destruction coordinator and providers and that they prefer to maintain a clear boundary between informal consultations (in which providers share information voluntarily) and formal investigations (which may involve subpoenaed information from providers). As noted in Recommendation 4, the threshold of concern above which sharing is required should be tuned in response to the burdens of screening and changes in bioengineering capabilities.

5. Require red-flagged orders to have a second approver from the same institution, such as a biosafety officer, mitigating some accident risks and insider threats

Certain gene synthesis orders should not be filled without ensuring that the customer will conduct their work with appropriate biosafety (preventing accidental exposure to or release of dangerous agents) and biosecurity (preventing unauthorized access, theft, or misuse of dangerous agents) measures. Requiring a second approver to confirm that work with certain sequences of concern is authorized and will be performed with appropriate biosafety and biosecurity measures would address several risks:

• Laboratory-acquired infections due to inadequate biosafety. Some sequences of concern, such as those that enhance the pathogenicity of a transmissible respiratory virus, could create pathogens that accidentally infect researchers and heighten risks of further transmissions that could reach epidemic or pandemic scale or other severe risks.
• Theft of dangerous materials due to inadequate biosecurity. For example, some bad actors may attempt to acquire pathogens from laboratories that conduct legitimate research.
• Insider threats. Individuals with malicious intent may attempt to covertly use a laboratory to which they have authorized access, and many laboratory supervisors do not maintain fine-grained awareness of all experiments being conducted in their laboratory.

The second approver should be someone with knowledge of laboratory biorisks and awareness of the customer’s work. In many academic contexts, this should be the biosafety officer overseeing the work, as it is their responsibility to ensure that work within their institution is conducted with appropriate biorisk management (e.g., the practices outlined in the US Biosafety in Microbiological and Biomedical Laboratories³⁵ manual and WHO Laboratory Biosafety Manual³⁶). In many industrial contexts, the second approver should be the Environmental Health and Safety professional overseeing laboratory worker safety. When a customer cannot identify an individual at their institution responsible for biosafety and biosecurity oversight, case-by-case determinations will be needed to identify appropriate second approvers.

A second approver system is already in place for some providers of physical materials. Before ATCC allows customers to purchase any materials other than simple chemical reagents, a biosafety officer must electronically sign the customer’s application for a purchasing account.³⁷ Ordering pathogens at high biosafety levels from BEI Resources, an infectious disease culture repository hosted under ATCC, requires extensive documentation.³⁸ However, contacts reported that this process can be cumbersome due to the need to print out documents on paper, sign them, and then hand them off to the biosafety officer for a second signature.³⁹

The risks from synthetic DNA are lower than those from active viral cultures, and orders are made more frequently, so this second approver system must be more scalable than existing processes. If the volume of orders for review is too high, second approvers may begin to rubber-stamp orders without a detailed review. This is why we recommend that a second approver system be used only for orders above a relatively high threshold of concern (which, as discussed under Recommendation 4, should be tuned over time in response to the screening burdens). Detailed review could be encouraged by providing a streamlined digital interface for order review and approval orders and including a recognition of liability during the approval process.

Conclusion

When securing nucleic acid synthesis, follow-up screening and reporting must be addressed. Many decisions about synthesis order fulfillment hinge on questions about the customer, specifically: Do they have a legitimate, peaceful purpose for the flagged sequences of concern?

Regulators have a rare opportunity to introduce future-proofed standards for record-keeping and information sharing that will reduce the burden on providers, introduce a consistent process, and lay the groundwork for further automation of screening.

For follow-up screening to keep pace with the industry’s growth, we recommend that the US government move quickly to pursue the steps we recommend in this briefer.

---

About the Authors

Tessa Alexanian is an Ending Bioweapons Fellow at the Council on Strategic Risks and Technical Lead, Common Mechanism at International Biosecurity and Biosafety Initiative for Science (IBBIS).

Sella Nevo is the Director of the Meselson Center at the RAND Corporation, a Venture Partner at Firstime VC, and an Advisory Board Member at the Alliance to Feed the Earth in Disasters.

---

Notes