EU · REG · 1223/2009 Consumer Advisory Est · MMXXVI
BANNED in Europe Banned in Europe
Explainers

The GRAS Loophole: How a Chemical Skips FDA Review

Most people assume the FDA signs off on every ingredient before it reaches your plate. For a whole category of additives, it doesn't have to. Here's how that happened, and what it actually does and doesn't prove.

By 8 min read Explainers

Here’s a thing most people believe that isn’t quite true: that the FDA checks every ingredient before it goes into your food. It’s a reasonable assumption. There’s an agency, there’s a process, there are forms. Surely someone looked.

For a large and growing category of additives, no one at the FDA did. They didn’t review it, didn’t approve it, and in many cases don’t even know it’s in the food supply. The mechanism that allows this has a wonky, almost soothing name: GRAS, short for “generally recognized as safe.” And it’s one of those stories where the outrage is half right, which is the most interesting kind.

So let me lay out how it actually works, because the real problem isn’t the one the scary headlines usually name.

Where GRAS came from, and why it made sense

Start in 1958. Congress passed the Food Additives Amendment, which President Eisenhower signed, and it did something genuinely useful: it required companies to prove a new food additive was safe before selling it, and to get FDA sign-off first. Premarket approval. That was the headline.

But Congress carved out an exception, and the exception was sensible at the time. It didn’t make sense to put salt, vinegar, baking soda, and vegetable oil through a full toxicology review. These were substances with centuries of safe use, things “generally recognized, among experts qualified by scientific training and experience,” as safe under their intended conditions. So the law exempted them from the “food additive” definition and the approval process that came with it. (FDA, history of the GRAS provision; 21 U.S.C. via FDA overview)

That’s the original intent: a common-sense lane for the obviously benign. Salt doesn’t need a clinical trial. Hard to argue with.

The trouble is what the lane became.

The 1997 turn: from “tell us” to “tell us if you feel like it”

For a while, if a company thought a new ingredient qualified as GRAS, it could petition the FDA to formally affirm that. The agency would review the petition and publish a decision. Slow, but on the record.

In 1997, the FDA changed course. It proposed replacing the affirmation petitions with a voluntary notification system. The agency had essentially concluded it could no longer pour resources into the time-consuming affirmation process. Under the new approach, a company could notify the FDA that it had determined a substance was GRAS, the FDA would review the notice and usually respond with a “no questions” letter, and everyone moved on. The FDA finalized this as a formal rule in August 2016. (Federal Register, GRAS final rule, 81 FR 54960, Aug. 17, 2016; FDA, about the GRAS notification program)

Read that word again: voluntary. A company can determine, on its own, that an ingredient is GRAS, and then it can choose not to tell the FDA at all. It can convene its own panel of experts, review its own data, write its own conclusion, and start adding the substance to food. No filing. No public record. No “no questions” letter, because no question was ever asked.

This is the loophole. It isn’t that GRAS exists. It’s that “generally recognized as safe” quietly came to include “recognized as safe by the company that wants to sell it, in private.”

How many ingredients are we talking about

This is where the numbers get slippery, so let me be careful.

The most-cited figure comes from the Natural Resources Defense Council, which estimated that since 1958, roughly 1,000 chemicals have entered the US food supply through GRAS determinations that the FDA was never told about. Not reviewed, not notified, in some cases known only to the manufacturer. (NRDC, “Generally Recognized as Secret”)

You’ll also see a bigger, scarier number floating around: that something like 10,000 additives are allowed in US food. That one is real, and it comes from the Pew Charitable Trusts, which reviewed the whole system around 2013. Pew’s breakdown is the useful part. Of roughly 10,000 permitted chemicals (about half used directly in food, the rest in packaging and other food-contact materials), the FDA had never reviewed the safety of about 3,000. And of those, roughly 1,000 were self-affirmed as GRAS by manufacturers without the agency ever being told. (Pew Charitable Trusts, “Fixing the Oversight of Chemicals Added to Our Food,” 2013)

So the “10,000 chemicals” headline isn’t 10,000 secret poisons. It’s the whole permitted universe, most of it reviewed. The genuinely-never-seen-by-FDA slice is the ~1,000, and even that is an estimate, precisely because the whole point of the loophole is that nobody has to count.

The fact that the best available number is an educated guess is, itself, the most damning detail in the story.

What this does not prove

Here’s the part the viral version skips.

“GRAS” does not mean “untested,” and it definitely doesn’t mean “poison.” A substance can be self-affirmed GRAS and still be backed by a thick stack of legitimate safety data. Plenty of GRAS ingredients are exactly what the 1958 lawmakers had in mind: well-understood, low-risk, boring in the best way. The category includes things no sane person worries about.

It’s also worth saying that the voluntary notification program does real work. The FDA has reviewed hundreds of GRAS notices since the first one arrived in 1998, and it does sometimes push back, ask for more data, or decline to issue a “no questions” letter. The system isn’t a pure rubber stamp. (FDA GRAS notice inventory)

So if someone tells you every GRAS ingredient is a hidden toxin, they’re overselling it in the same way the “European food is pure” crowd oversells the other direction. The substance of the critique is narrower, and stronger for being narrow.

What’s actually wrong with it

Two things, mainly.

The first is secrecy. When a company self-determines GRAS and skips notification, there’s no public record that the ingredient is in use, no central list, and no way for an outside scientist, journalist, or regulator to know it exists until they find it on a label and reverse-engineer what it’s doing there. You can’t study, recall, or even debate a chemical you don’t know is in the food supply.

The second is conflict of interest. The company picks the experts who judge its own ingredient. A 2013 analysis in JAMA Internal Medicine by Neltner and colleagues examined 451 GRAS determinations submitted to the FDA between 1997 and 2012 and found that every single one was made by someone with a financial tie to the manufacturer. In 22.4% of cases, the safety judgment came from an employee of the very company making the additive; the rest went to consultants or expert panels the manufacturer chose and paid. Across the 290 expert panels in the review, 100% of the members worked, directly or indirectly, for the company whose ingredient they were judging. (Neltner et al., JAMA Internal Medicine, 2013) If you designed a system to maximize the appearance of independence while minimizing the reality of it, you’d land somewhere close to this.

There’s movement, at least. In March 2025, the Secretary of Health and Human Services directed the FDA to explore rulemaking that would eliminate the self-affirmed pathway and require companies to notify the agency, with their safety data, before a new ingredient reaches the food supply. (HHS press release, March 10, 2025) As of mid-2026 that’s a stated intention working its way toward a proposed rule, not a finished regulation. Worth watching, not yet worth celebrating.

The European contrast, and why it’s the real headline

This is where the site’s usual theme shows up cleanly. The EU has no equivalent of self-affirmed GRAS. Food additives there run through a positive-list system: a substance has to be evaluated, usually by the European Food Safety Authority, and explicitly authorized and listed before it can be used, under Regulation (EC) No 1333/2008. (Regulation (EC) No 1333/2008, EUR-Lex) The default is “not allowed until cleared,” and the clearing is done by a public body, not the seller.

That’s the same precaution-versus-permission split you see across EU and US chemical rules more broadly, and it’s the honest version of the “feel better in Europe” instinct. The difference isn’t that European food is magically clean. It’s that European regulators insist on seeing the homework before the test, while the US system, for a big slice of ingredients, lets companies grade their own.

One last illustration, because it shows the loophole has teeth. Partially hydrogenated oils, the source of artificial trans fats, were once treated as GRAS. The evidence eventually became overwhelming that they weren’t safe, and in 2015 the FDA revoked their GRAS status, with a compliance phase-out following. (FDA on partially hydrogenated oils) A substance the system once “generally recognized as safe” turned out to cause tens of thousands of heart attacks. The loophole isn’t theoretical. It just usually operates in the dark, which is the whole problem.

The bottom line

GRAS started as a sensible exemption for salt and vinegar and drifted into a pathway where a company can add a novel chemical to your food, decide for itself that it’s safe, and never tell anyone. Most GRAS ingredients are probably fine. A meaningful number have never been independently reviewed, and nobody can tell you exactly how many, because the system was built so they wouldn’t have to. That’s not a reason to panic about your pantry. It’s a reason to want the rule changed.

Sources