This is the first of a series of posts on scholarship related to food and drug law posted on SSRN in the first half of 2026. The two pieces below share a topic — the significant difference between regulation as a food and regulation as a drug — and circle around a common question: why do we ask certain things (before market entry) of the one, and not the other, and does this always make sense?
(Subtitle: A Comprehensive Proposal to Reshape Food Safety Review by Treating Food as Medicine.)
The title of this piece, authored by Katya Cronin of GW Law and published in 2025 in the American Journal of Law & Medicine, gave me the (incorrect) impression it would talk about the blurred line between foods and drugs and perhaps even propose reclassification of some (or all) food as medicine. Instead, it is very solidly grounded in the real world — with a succinct and clear description of current regulatory requirements with respect to the safety of food ingredients [which she defines to include unintended environmental contaminants], and then detailed suggestions for both Congress, and FDA, to make the food regulatory paradigm much more rigorous with respect to ingredient safety.
What Congress should do. Ultimately, she argues that (1) Congress should eliminate the GRAS carve-out from the food additive definition, or in the alternative limit its application to substances that have at least twenty years of use, and at the very least mandate submission of GRAS certifications to the agency; (2) eliminate the food contact notification program in favor of a full premarket petition process and clarify that the Delaney Clause applies with equal force, and without de minimis exception, to GRAS substances as well as all direct and indirect food additives and food contact substances; (3) enact user fees for all food-related premarket submissions; (4) require clinical studies for any new food or color additive; and (5) authorize FDA to require more extensive reporting from food manufacturers.
What FDA should so. In the meantime, and even if Congress fails to act, she argues that FDA can and should take numerous steps on its own. For instance, she says, it should convert the voluntary GRAS notification system to a mandatory notification system and issue guidance stating that novel substances cannot be subject to GRAS determinations. She also argues that the agency should “enforce in practice the manufacturers’ legal burden of proving safety.” Thus, she says, it should “examine all filed GRAS notices and approved food and color additives and should summarily revoke those that are filed without sufficient toxicology, feeding toxicity, and other relevant data.” It should use artificial intelligence to identify approved additives as to which there are “reputable, peer-reviewed studies with adverse safety results,” she says, arguing that “the very existence of such studies should sufficiently demonstrate that the manufacturer failed to demonstrate reasonable certainty of safety.” Third, she says, FDA should require food manufacturers to test “every lot” for heavy metals, environmental toxins, and pathogens. And finally, she argues that FDA should require that all food and color additives be listed on food labels and that the manufacturer’s labeling information (on their website or otherwise available) contain data on the relevant safety testing that supports the determination that the ingredient is safe for use.
Striking while the iron is hot? Most of these proposals are not, I think, new to food policy circles, but they are helpfully gathered here and stated clearly and precisely. I think there is more to say about the proposals specific to FDA — in terms of resource availability and options, as well as potential legal constraints — but that would be (perhaps is?) another article. This is a helpful “gather everything together and start the conversation” piece. And she has framed this to be exceptionally timely, borrowing from the “food is medicine” zeitgeist (and the observation that “medicine” has much more rigorous safety requirements), and pointing to political momentum to reform the food system.
As the next piece suggests, though, a more drug-like paradigm would have its costs…
This one, also, was not about the blurred line between food and drugs! (I’ve been thinking about Lewis Grossman’s article, Food, Drugs, and Droods, as well as FDA’s famous warning letter to General Mills about its Cheerios being a new drug on account of something printed on the cereal box.)
A new animal drug application … for my grilled salmon. Instead, this article — written by Hadar David, a doctoral candidate at Stanford Law School — concerns itself with genetically engineered food animals, such as AquAdvantage Salmon, which was genetically engineered to grow more quickly than its Atlantic salmon counterpart and approved by FDA in 2015. And to be clear about the regulatory treatment: this required approval of a new animal drug application (NADA). David writes, “although food products from GE animals are intended for consumption as food, they are regulated not as food but as drugs, triggering a premarket approval process that involves rigorous and time-intensive data requirements.” (And, both are true. Genetically modifying the salmon in the first instance requires approval of an NADA, but the food product made from the salmon is still regulated as a food. Thus, as FDA explained in a Q&A at the time, USDA would regulate the disclosure of bioengineered content on the labeling of any human food made from the GE salmon.)
Bad for innovation. Part III of David’s article argues that application of the drug paradigm to GE food animals “has had a profound impact on both industry and academic stakeholders.” Specifically, he says, it has had two deleterious results: (1) “the complexity, cost, and duration of the FDA’s drug approval process have significantly narrowed the field of actors willing or able to engage in GE food animal development,” and (2) “the scope of innovation has been curtailed: regulatory burdens have led to a concentration on a limited number of commercially viable traits, while more experimental or socially beneficial applications remain unexplored.”
Is it really a “drug”? A short section in the middle of this article evaluates FDA’s classification of genetic modifications in GE food animals as “drugs.” David concedes that courts are unlikely to take into account the policy implications of the classification. And he notes there are questions about who could (and would) seek judicial review. But he nevertheless takes up the basic issue, making an argument from the statutory text and legislative purpose that the agency’s interpretation is problematic. In the end, this argument is unlikely to prevail, and I think he realizes that. In 2019 — as he notes — a district court in the N.D. California upheld FDA’s authority to regulate the (GE salmon) product. This was Institute for Fisheries Authorities v. Hahn, 424 F. Supp. 3d 740 (N.D. Cal. 2019). The court agreed that FDA can treat the recombinant DNA “construct” used to modify the animal as a “drug” because it is intended to affect the structure or function of the animal. It may also be worth noting that the court reached this conclusion without applying the Chevron framework; i.e., it decided without much fanfare that the plain language of the statute answered the question — placing the construct within the term “drug.” Consequently there is no reason to think a court would come out differently now, with Chevron overruled. (And I think some of the arguments in that case were similar to those David offers.)
In search of a better approach. But the heart of David’s argument is more philosophical. He’s concerned about the policy implications of the classification, and I think also bothered by the basic conceptual difference between modifying an animal for therapeutic purposes and modifying it for food production purposes. (One could devote an entire article to this — going back to the old question of how to cabin the structure/function prong of the drug and device definitions — but that is a different article.) In any case, he argues that the complexity of the subject, and the need to consider the broader consequences of the classification chosen, mean the courts are not the right place to turn for a coherent long-term framework. To be sure!
Section V then discusses non-drug alternatives (as a food? as a food additive?) before settling on new legislation, which he proposes would take lessons from Argentina and Brazil in particular. I trust, and hope, this can be fleshed out in a subsequent article in greater detail, but the idea seems to be making some basic distinctions — such as between transgenic modifications (introducing genes from another organism) and non-transgenic alterations (such as knock-out situations), though he notes that this categorical distinction might not correspond to actual risk. It would be very interesting to see this fleshed out, perhaps in another article.