The Need for Stronger Regulation of Engineered Live Microbes as Foods 

By Benjamin Snitkoff, JD, MS and Frank S. David, MD, PhD

Biotechnology has enabled the development of genetically modified non-pathogenic microbial strains intended for human use. Some of these engineered live microbes (ELMs) that do not meet the U.S. Food and Drug Administration’s (FDA’s) definition of a drug can reach the market under the food regulatory regime, which allows them to take advantage of a significantly lower level of scrutiny. In particular, if a substance intended for a food use is “generally recognized as safe” (GRAS), it is exempt from the food additive approval process. The FDA previously took the position that it lacks express statutory authority to require companies to submit notices to FDA that a given substance has been found to be GRAS, and accordingly allows manufacturers to self-affirm food products as GRAS. This allows companies to bypass the standards for demonstrating safety under the food additive approval process.[1],[2] At least two ELMs are currently sold directly to consumers as foods designated as GRAS by their manufacturer.

We argue here that allowing ELMs to reach the market as foods using the low-stringency self-affirmed GRAS pathway is inappropriate given the potential risks these products pose to the user, other individuals, and the environment. In general, there are no salient scientific distinctions between these products and products that are subjected to stricter scrutiny as bona fide drugs that would make them inherently safer. Thus, ELMs’ marketing approval as foods should require more formal FDA oversight.

Importantly, although there is a debate underway at the time of writing this article over the fate of the self-affirmed GRAS pathway as a whole, our focus here is on the narrower question of whether ELMs should be allowed to utilize this path. We argue here that it is appropriate, warranted, and feasible in the near-term to exclude ELMs from self-affirmed GRAS, without waiting for broader questions about the GRAS program’s fate to be resolved.

Regulatory Regimes

The FDA defines a substance as a drug if it is “intended for use in the diagnosis, cure, mitigation, treatment, or prevention of disease in man or other animals.”[3],[4] Drugs are evaluated for safety and efficacy both before and after approval through preclinical and clinical testing, ongoing post-approval monitoring, and mandatory reporting of adverse events and long-term patient outcomes.[5] Even once a drug is approved and enters the market, the FDA can require changes to its label or revoke its approval based on new data.

In contrast, new substances that fall outside the drug definition and are added to foods or may contact foods through packaging are regulated by the FDA via one of two pathways. Under the more stringent food additive path, approval involves engaging in pre-petition consultations with the FDA; assembling data on chemical composition, manufacturing, purification processes, and toxicological and safety information; and submitting a formal Food Additive Petition after input from and discussions with the agency. The petition is then open for public comment before approval. If approved, the FDA would publish the approval in the U.S. Code of Regulations (the CFR), which may include relevant requirements on labeling and approved uses.[6]

Alternatively, a manufacturer of a new food product may opt for the less stringent GRAS designation if the substance is “generally recognized, among experts qualified by scientific training and experience to evaluate its safety, as having been adequately shown through scientific procedures to be safe.”[7] A company may voluntarily file a “notification” with the FDA to explain its determination that a new product qualifies as GRAS, in response to which the agency can issue a “no questions” letter if it agrees with (or at least does not object to) the manufacturer’s assessment. However, a firm can also choose the more lenient option of self-affirmation, which does not require notice to or input from the FDA.[8] Although the exact number of new self-affirmed GRAS products that reach the market is unknown due to the lack of reporting requirements, available information strongly suggests that this pathway is used much more frequently than GRAS notification.[9]

The self-affirmed GRAS pathway has been subject to extensive criticism over its lax requirements and lack of transparency. Many academics have argued for its elimination,[10],[11],[12], [13],[14] and the U.S. Secretary of Health and Human Services recently directed the FDA to explore this possibility.[15]

Regulatory Paths Taken by Engineered Live Microbes to Date

ELMs intended for human therapeutic use, commonly referred to as Live Biotherapeutic Products (LBPs) by regulators and academics, are covered by drug regulations. To date, none have been approved as drugs by the FDA, but several have been tested in clinical trials for a range of uses.[16] These agents are subject to FDA requirements for pre- and post-approval testing like all other prescription medicines, with some additional specific elements.[17]

On the food side, we are aware of two ELMs that are currently on the market and explicitly avoid making any therapeutic claims. Pre-Alcohol (ZB183, ZBiotics) is a Bacillus subtilis strain engineered to secrete an exogenous acetaldehyde dehydrogenase. The manufacturer’s website says it “[b]reaks down a toxic byproduct of alcohol called acetaldehyde while you drink and while you sleep – setting you up for a great next morning.”[18] Sugar-to-Fiber (ZB423, ZBiotics) is a Bacillus subtilis strain engineered to constitutively express an endogenous levansucrase. According to the company, the enzyme “makes fiber using sugar from your diet, gently and all day long–improving microbiome health.”[19]

ZBiotics appears to have used the GRAS self-affirmation process for both products, based on their absence from the FDA’s online database of GRAS notification filings.[20] Although the company was not required to publicly state its reasoning, one can surmise that its rationale relied at least in part on the fact that the products were derived from well-characterized strains that are recognized for their tolerability by humans.

For completeness, we also note that at least one ELM appears to have reached the U.S. market outside of both food and drug regulatory regimes.[21] Lantern Bioworks’ Lumina Probiotic is a strain of the oral bacterium Streptococcus mutans genetically engineered to metabolize ingested food into ethanol instead of lactic acid.[22] Because the product is explicitly characterized by the company as a cosmetic,[23] we consider it outside this scope of this analysis.

Types of Real and Potential Risks Posed by ELMs

Genetic engineering confers potential risks on ELMs that may exceed those posed by unmodified live microbial food substances, such as those found in cultured dairy and fermented products or genetically engineered yeasts that are killed and filtered from alcoholic beverages before packaging. These potential risks fall into three categories: risks to the user, risks to others, and risks to the environment.

Risks to the user. Although many microbial products for human use have the potential to colonize the user, at least transiently,[24] manufacturers of ELMs pursuing marketing under the self-affirmed GRAS pathway are not required to generate relevant safety data for review by FDA. This stands in contrast to pharmaceutical companies seeking approval of LBPs through the drug pathway, which typically assess the risk of colonization and alteration of the user’s microbiome and take proactive steps to mitigate it.

Several manufacturers of LBPs intended for therapeutic use have performed rigorous safety testing and used genetic engineering to reduce risks to users. Synlogic tested SYNB1934, its LBP drug candidate for phenylketonuria, to understand how long it remained in a patient after dosing, even though the parental E. coli strain was already known to not permanently establish itself in its host.[25] SYNB1934 was also engineered to be auxotrophic for (i.e., dependent for growth on) diaminopimelate, a necessary cell-wall component, to limit its ability to reproduce and colonize the gut.[26] Similarly, Aurealis Therapeutics’s AUP-1602C (also referred to as AUP-16), a modified Lactococcus cremoris bacterium in clinical trials for diabetic foot ulcers, has an engineered auxotrophy that renders it dependent on exogenous D-alanine[27] that it can only obtain from growth medium or cross-feeding from other bacteria.[28] Another approach is to use auxotrophy to permit long-term but reversible colonization: an oxalate-degrading Phocaeicola vulgatus strain designed by Novome Biotechnologies (now defunct) to reduce kidney stone formation was also engineered to depend on exogenously supplied porphyrin, a carbohydrate found in red algae, for growth.[29]

An ELM could also alter the microbiome of a user through horizontal transfer of genetic material to naturally occurring bacteria in the user’s microbiome. There is evidence of horizontal gene transfer in the human gut,[30] including antibiotic resistance genes,[31] which are commonly used in genetic engineering. FDA guidance on LBPs being tested under the drug regime specifically requests data on antibiotic resistance and whether such resistance can be transferred to endogenous microbial flora.[32] Although ZBiotics’s website says its products show no “transferable or extrinsic antibiotic resistance,” the underlying data have not been made public, and there is no evidence they have been reviewed by regulators.[33]

Risks to others. The transmission of human-associated microbiota has been well established between mothers and their children, between spouses, and within communities.[34],[35]Thus, it is theoretically possible for the user of an ELM to unknowingly transfer the organism to a third party without either’s knowledge or consent. The National Institutes of Health’s Recombinant DNA Advisory Committee (RAC) requested additional clinical safety studies to characterize these risks when it evaluated the aforementioned variant of Streptococcus mutans engineered to reduce oral acid production and subsequent tooth decay.[36]

Risks to the environment. Under the National Environmental Policy Act of 1969, all federal agencies are required to take environmental impact into account when evaluating proposed actions.[37] The FDA issued guidance in 1998 on integrating environmental assessments into the evaluation of applications for new drugs and biologics,[38] but this document does not explicitly address ELMs, and there are no equivalent guidelines for food-related reviews.

Even without establishing themselves in a host, ELMs pose a theoretical risk of environmental contamination, given that they are not naturally occurring and many can freely reproduce. ELMs may also participate in horizontal gene transfer in the wild, sharing their engineered genes and antibiotic resistance with naturally occurring bacteria. The Bacillus subtilis strains of the type used by ZBiotics form spores that are strongly heat and desiccation resistant, giving the bacteria the ability to survive in the environment indefinitely.[39] The choice of a spore-forming bacterium makes the product shelf-stable and easy to transport, but may also increase the risk of environmental contamination.[40] Engineered auxotrophies and “kill switches” can significantly mitigate environmental risk.

Potential Risk Mitigation Strategy for ELMs Regulated as Foods

Because of their potentially wide spectrum of risk, ELMs as foods should be reviewed by FDA on a case-by-case basis, and FDA should have the flexibility to require additional data or safeguards as appropriate. Although some ELMs may be suitable for a relatively low level of regulatory burden, others may warrant a higher degree of scrutiny—more akin to what is required for engineered LBPs regulated as drugs. Thus, given the uncertainty about the risks of products in this category, we believe ELMs should not be eligible for self-affirmed GRAS.

Short of eliminating self-affirmed GRAS entirely—which, as noted earlier, is currently under discussion—a possible solution would be for the FDA to amend GRAS regulations through notice-and-comment rulemaking to prevent ELMs from reaching the market solely on the basis of undisclosed manufacturer claims. This process, though arduous and likely to face legal challenges, could lead to mandatory formal premarket review of food-related ELMs, which would allow the agency to evaluate the safety considerations detailed above on a case-by-case basis and recommend or require key safeguards.

Another option to bring food-related ELMs under a more appropriate and tighter level of regulatory scrutiny would be for Congress to pass an amendment to 21 U.S.C. § 321(s) adding genetically engineered live microbes to the list of products that require formal definition as food additives, thus excluding them from the GRAS pathway. Though even more challenging than rulemaking, this approach would avoid questions about the scope of the FDA’s ability to make substantive changes via regulation alone in the aftermath of the Supreme Court’s 2024 ruling in Loper Bright Enterprises v. Raimondo.[41]

Conclusion

We maintain that the FDA needs to comprehensively evaluate the safety of ELMs developed as foods or food additives and have the flexibility to define the appropriate level of risk mitigation that these products need to demonstrate to reach the public. As long as the self-affirmed GRAS pathway remains a regulatory option for food manufacturers, we believe ELMs should be excluded from utilizing it and instead required to provide additional safety assurances before being formally authorized.

Acknowledgements

We gratefully acknowledge helpful comments from Emily Broad Leib, Lewis Grossman, Michael Sinha, Diana Winters, and Benjamin Wolfe on an earlier version of this manuscript, as well as helpful comments from an anonymous reviewer of this submission. This work was not funded. We acknowledge the use of Artificial Intelligence (AI) assistance in the writing process of this manuscript. The AI tool used was ChatGPT 5.4, which contributed to argument refinement. The final manuscript represents our own ideas and interpretations.

Conflict Statement

BS is employed by Ginkgo Bioworks, Inc., a company operating in the synthetic biology field, though Ginkgo does not manufacture or market ELMs. The views expressed are the authors’ own and do not necessarily reflect those of Ginkgo Bioworks. FSD is Managing Director of Pharmagellan, a consulting firm that serves as a paid advisor to numerous biotechnology companies and investors; no companies mentioned in this article are current or past clients of the firm.

[1] U.S. Food and Drug Administration, FDA’s Approach to the GRAS Provision: A History of Processes, https://www.fda.gov/food/generally-recognized-safe-gras/fdas-approach-gras-provision-history-processes (accessed Feb. 2, 2026).

[2] Constituent Update, U.S. Food and Drug Administration, FDA Issues Final Rule on Food Ingredients that May Be “Generally Recognized as Safe” (Aug. 12, 2016), https://www.fda.gov/food/hfp-constituent-updates/fda-issues-final-rule-food-ingredients-may-be-generally-recognized-safe.

[3] 21 U.S.C. § 321(g)(1).

[4] U.S. Food and Drug Administration, Small Entity Compliance Guide on Structure/Function Claims, https://www.fda.gov/regulatory-information/search-fda-guidance-documents/small-entity-compliance-guide-structurefunction-claims (accessed Feb. 2, 2026).

[5] Congressional Research Service, How FDA Approves Drugs and Regulates Their Safety and Effectiveness (CRS Report No. R41983), May 8, 2018.

[6] Magali Cordaillat-Simmons, Alice Rouanet & Bruno Pot, Live biotherapeutic products: the importance of a defined regulatory framework, 52 Exp. Mol. Med. 1397 (2020).

[7] 21 U.S.C. § 321(s).

[8] U.S. Food and Drug Administration, FDA’s Approach to the GRAS Provision: A History of Processes, https://www.fda.gov/food/generally-recognized-safe-gras/fdas-approach-gras-provision-history-processes (accessed Feb. 2, 2026).

[9] Pew Charitable Trusts, Fixing the Oversight of Chemicals Added to Our Food, Nov. 7, 2013.

[10] Jennifer L. Pomeranz, Emily M. Broad Lieb & Dariush Mozaffarian, Regulation of added substances in the food supply by the Food and Drug Administration human foods program, 114 Am. J. Pub. Health 1061 (2024).

[11] Pieter A. Cohen & Emily M. Broad Leib, Ingesting risk — The FDA and new food ingredients, 391 N. Eng. J. Med. 875 (2024).

[12] Martha Dragich, GRAS-Fed Americans: Sick of Lax Regulation of Food Additives, 49 Ind. L. Rev. 305 (2015)

[13] Frederick H. Degnan, Rethinking the Applicability and Usefulness of the GRAS Concept, 46 Food Drug Cosmetic L. J. 553 (1991).

[14] Roxana R. Soroudi, Government Repudiation of Americans’ Safety: A Call for Reformulation of FDA’s GRAS Notification Program, 75 Food & Drug L. J. 39 (2020).

[15] Press Release, U.S. Department of Health and Human Services, HHS Secretary Kennedy Directs FDA to Explore Rulemaking to Eliminate Pathway for Companies to Self-Affirm Food Ingredients Are Safe (Mar. 10, 2025), https://www.hhs.gov/press-room/revising-gras-pathway.html.

[16] Rajkamal Srivastava & Cammie F. Lesser, Living Engineered Bacterial Therapeutics: Emerging Affordable Precision Interventions, 17 Microb. Biotechnol. e70057 (2024).

[17] U.S. Food and Drug Administration, Early Clinical Trials with Live Biotherapeutic Products: Chemistry, Manufacturing, and Control Information; Guidance for Industry (June 2016), https://www.fda.gov/regulatory-information/search-fda-guidance-documents/early-clinical-trials-live-biotherapeutic-products-chemistry-manufacturing-and-control-information.

[18] Home page, ZBiotics, https://zbiotics.com (accessed Feb. 2, 2026).

[19] Id.

[20] U.S. Food and Drug Administration, GRAS Notices, https://www.hfpappexternal.fda.gov/scripts/fdcc/index.cfm?set=GRASNotices (accessed Feb. 2, 2026).

[21] Home page, Lumina Probiotic, https://luminaprobiotic.com/ [https://perma.cc/L275-JWA6] (accessed Feb. 2, 2026).

[22] Jeffrey D. Hillman, Genetically modified Streptococcus mutans for the prevention of dental caries, 82 Antonie Van Leeuwenhoek 361 (2002).

[23] Lumina Probiotic, supra note 21.

[24] Daniel Merenstein, Bruno Pot, Gregory Leyer, Arthur C. Ouwehand, Geoffrey A. Preidis, Christopher A. Elkins, Colin Hill, Zachery T. Lewis, Andi L. Shane, Niv Zmora, Mariya I. Petrova, Maria Carmen Collado, Lorenzo Morelli, Gina A. Montoya, Hania Szajewska, Daniel J. Tancredi & Mary Ellen Sanders, Emerging issues in probiotic safety: 2023 perspectives, 15 Gut Microbes 2185034 (2023).

[25] Marja K. Puurunen, Jerry Vockley, Shawn L. Searle, Stephanie J. Sacharow, John A. Phillips 3rd, William S. Denney, Benjamin D. Goodlett, David A. Wagner, Larry Blankstein, Mary J. Castillo, Mark R. Charbonneau, Vincent M. Isabella, Vasu V. Sethuraman, Richard J. Riese, Caroline B. Kurtz & Aoife M Brennan, Safety and pharmacodynamics of an engineered E. coli Nissle for the treatment of phenylketonuria: a first-in-human phase 1/2a study, 3 Nature Metabolism 1125–32 (2021).

[26] Jerry Vockley, Neal Sondheimer, Marja Puurunen, George A. Diaz, Ilona Ginevic, Dorothy K. Grange, Cary Harding, Hope Northrup, John A. Phillips 3rd, Shawn Searle, Janet A. Thomas, Roberto Zori, William S. Denney, Sharon L. Ernst, Kristina Humphreys, Nicole McWhorter, Caroline Kurtz & Aoife M. Brennan, Efficacy and safety of a synthetic biotic for treatment of phenylketonuria: a phase 2 clinical trial, 5 Nature Metabolism 1685–90 (2023).

[27] Siamond Hosseini, Alex Curilovs & Simon M. Cutting, Biological Containment of Genetically Modified Bacillus subtilis, 84 Applied & Env’t Microbiology e02334-17 (2018).

[28] Christoph Schindler, Jacek Mikosiński, Pawel Mikosiński, Hanna-Riikka Kärkkäinen, Mirka Sanio, Jere Kurkipuro, Igor Mierau, Wesley Smith, Aki Vartiainen, Laurent Décory, Dirk Weber, Thomas Wirth, Juha Yrjänheikki, Sebastian Schellong & Haritha Samaranayake, Multi-target gene therapy AUP1602-C to improve healing and quality of life for diabetic foot ulcer patients: a phase I, open-label, dose-finding study, 15 Therapeutric Advances in Endocrinology & Metabolism 20420188241294134 (2024).

[29] Weston R. Whitaker, Zachary N. Russ, Elizabeth Stanley Shepherd, Lauren M. Popov, Alexander Louie, Kathy Lam, David M. Zong, Clare C. C. Gill, Jeanette L. Gehrig, Harneet S. Rishi, Jessica A. Tan, Areta Buness, Janeth Godoy, Domenique Banta, Sonia Jaidka, Katheryne Wilson, Jake Flood, Polina Bukshpun, Richard Yocum, David N. Cook, Tariq Warsi, Lachy McLean, Justin L. Sonnenburg & William C. DeLoache, Controlled colonization of the human gut with a genetically engineered microbial therapeutic, 389 Science 303–08 (2025).

[30] Tiffany Y. Xu, Etienne Nzabarushimana, Dennis Wong, Chengwei Luo, Robert G. Beiko, Morgan Langille, Curtis Huttenhower, Long H. Nguyen & Eric A. Franzosa, Profiling novel lateral gene transfer events in the human microbiome, bioRxiv 2023.08.08.552500 (2023).

[31] Alexander Crits-Christoph, Haley Anne Hallowell, Kalia Koutuovalis & Jotham Suez, Good microbes, bad genes? The dissemination of antimicrobial resistance in the human microbiome, 14 Gut Microbes 2055944 (2022).

[32] U.S. Food and Drug Administration, supra note 17.

[33] Product Safety, ZBiotics, https://zbiotics.com/blogs/journal/product-safety [https://perma.cc/3SN9-BSKS] (accessed Feb. 2, 2026).

[34] Katri Korpela, Paul Costea, Luis Pedro Coelho, Stefanie Kandels-Lewis, Gonneke Willemsen, Dorret I. Boomsma, Nicola Segata & Peer Bork, Selective maternal seeding and environment shape the human gut microbiome, 28 Genome Rsch. 561–68 (2018).

[35] Ilana L. Brito, Thomas Gurry, Shijie Zhao, Katherine Huang, Sarah K. Young, Terrence P. Shea, Waisea Naisilisili, Aaron P. Jenkins, Stacy D. Jupiter, Dirk Gevers & Eric J. Alm, Transmission of human-associated microbiota along family and social networks, 4 Nature Microbiology 964–71 (2019).

[36] National Institutes of Health, Recombinant DNA Advisory Committee, Minutes of Meeting, March 9–11, 2004, https://web.archive.org/web/20041030140923/http://www4.od.nih.gov/oba/rac/minutes/RAC_minutes_03-04.pdf (accessed Feb. 2, 2026).

[37] 42 U.S.C. §§ 4321–4347.

[38] U.S. Food and Drug Administration, Environmental Assessment of Human Drug and Biologics Applications, https://www.fda.gov/regulatory-information/search-fda-guidance-documents/environmental-assessment-human-drug-and-biologics-applications (accessed Feb. 20, 2026).

[39] Peter Tuddenham, National Institutes of Health RAC Meeting March 10, 2004 (Vimeo, July 28, 2024), https://vimeo.com/991167628 (accessed Feb. 2, 2026).

[40] Lars Mølbak, Molin Søren & Niels Kroer, Root growth and exudate production define the frequency of horizontal plasmid transfer in the Rhizosphere, 59 FEMS Microbiology Ecology 167–76 (2007).

[41] David A. Simon & Michael J. Young, The Supreme Court’s Loper Bright Ruling: Implications for Clinical Testing, Innovation, and Public Health, 332 JAMA 1325 (2024).

 

FDLI is a nonprofit membership organization that offers education, training, publications, and professional networking opportunities in the field of food and drug law. As a neutral convener, FDLI provides a venue for stakeholders to inform innovative public policy, law, and regulation. Articles and any other material published in Update represent the opinions of the author(s) and should not be construed to reflect the opinions of FDLI, its staff, or its members. The factual accuracy of all statements in the articles and other materials is the sole responsibility of the authors.