GRAS: How an Ingredient Gets Into Your Food Without Anyone Approving It
If you understand this one concept, every ingredient label in the store reads differently.
There's a question that sits underneath every file we publish, and most people never think to ask it. Not is this ingredient safe — the question before that one:
Who decided?
When you pick up a package with thirty-one ingredients on the back, your brain fills in an assumption: somebody in a government building looked at each of these, ran the tests, signed off, and cleared it for your family. It's a reasonable thing to assume. It's how most people picture the system working.
For a large number of the substances in American food, that is not what happened. In many cases, nobody at the FDA reviewed the ingredient at all. In some cases, the FDA does not know the ingredient is in the food supply.
The mechanism that allows this has a name, and it's one of the least understood three-letter words in the American pantry: GRAS — Generally Recognized As Safe.
This is the file on how it works. What GRAS actually means, why Congress created it for genuinely sensible reasons, how it turned into something its authors likely didn't intend, what the FDA's "no questions" letter really says, and — because we don't publish haunted houses — the honest case that the system is not as broken as this article is about to make it sound.
Our tags as always. ESTABLISHED means settled and citable. CONTESTED means qualified people disagree. UNKNOWN means nobody's done the work. We don't tell you it's poison. We don't tell you it's fine. We hand you the file.
Start with the problem Congress was actually solving
To understand GRAS, you have to understand that it began as a reasonable idea, because almost every regulatory loophole in this library began as a reasonable idea.
In 1958, Congress passed the Food Additives Amendment — the law that, for the first time, required manufacturers to prove a new food additive was safe before putting it on the market. Before that, the burden ran the other way: the government had to prove something was harmful after it was already in your food. The 1958 law flipped that, and it was a genuine advance in consumer protection. [ESTABLISHED]
But a law like that creates an immediate practical problem. If every substance added to food requires a full pre-market safety review, then vinegar requires a safety review. Salt requires a safety review. Black pepper, baking soda, yeast, and every spice on the rack require dossiers, studies, and federal sign-off — despite humanity having eaten them, in quantity, for thousands of years without incident.
That would be absurd, so Congress carved out an exception. Substances generally recognized as safe by qualified experts — either through long common use in food before 1958, or through published scientific evidence — would not be classified as "food additives" and would not require pre-market approval. [ESTABLISHED]
Read that carve-out in its original context and it's hard to object to. It was meant to be a common-sense exemption for the obvious: don't make anyone file paperwork for cinnamon.
The question this whole article turns on is what happens when a carve-out written for cinnamon gets used for a novel synthetic compound that no human ate before 2015.
The part that surprises people: companies can decide for themselves
Here is the mechanism, stated as plainly as we can.
Under the GRAS pathway, a manufacturer can conclude on its own that a substance is generally recognized as safe, and begin using it in food, without submitting anything to the FDA. No application. No approval. In the fullest version of this — usually called self-affirmed GRAS or GRAS self-determination — the company assembles its own evidence, typically convenes its own panel of experts to review it, receives that panel's conclusion, files the paperwork in its own records, and goes to market. [ESTABLISHED]
The company may voluntarily notify the FDA. Many do. But notification is not required, and a company that skips it is not breaking the law. [ESTABLISHED]
The practical consequence is the fact that most people find genuinely startling when they first hear it: the FDA does not have a complete list of the substances in the American food supply. It cannot, structurally, because a substance can enter that supply through a determination the agency was never told about. Estimates of how many substances are in food through unreported self-determination have run to around a thousand, though by the nature of the problem no one can produce an exact count — the unknown quantity is the point. [ESTABLISHED that the gap exists; the specific number is an estimate — verify against the current Pew/GAO analyses before quoting a figure.]
This isn't a fringe reading or an activist framing. The Government Accountability Office — Congress's own investigative arm — examined this in a 2010 report and concluded that the FDA's oversight of GRAS substances was insufficient, specifically noting that the agency wasn't systematically reconsidering the safety of GRAS substances as new science emerged, and didn't have a full picture of what was in use. [ESTABLISHED — verify the report's exact findings; it's GAO-10-246.]
The sentence that isn't an approval
Now the part that matters most for reading labels and headlines, because this is where even careful people get misled.
When a company does voluntarily submit a GRAS notice, the FDA reviews it and, if it doesn't object, sends back a letter. That letter is widely described — in press releases, in marketing, sometimes in news coverage — as the FDA having "approved" or "cleared" the ingredient.
It says no such thing. The standard language is that the agency "has no questions at this time" regarding the company's conclusion that the substance is GRAS. [ESTABLISHED]
Sit with the construction of that sentence, because every word in it is load-bearing. The company's conclusion. The safety determination belongs to the manufacturer; the FDA is responding to it, not making it. No questions. Not "we verified this," not "we agree," not "we tested it." At this time. Explicitly provisional.
It is, in effect, the government declining to object — which is a meaningfully different act than the government affirming. A "no questions" letter is closer to a shrug than a stamp. [ESTABLISHED]
This distinction is not pedantry. It's the difference between the system most consumers believe exists and the one that does. And it's why, in our files, you'll see us repeatedly note that "the FDA approved it" is often not an accurate description of how a given substance got into your food.
Who is "generally recognized" doing the recognizing?
The phrase generally recognized as safe implies a consensus of disinterested experts. So the obvious question: who are they, and who picks them?
The answer is the structural weak point of the whole arrangement. In the self-determination model, the expert panel evaluating a company's ingredient is typically selected and paid by that company. [ESTABLISHED]
Researchers have examined how often this produces conflicts. Work published in the medical literature — notably analyses by researchers associated with the Pew Charitable Trusts — reviewed GRAS determinations and found financial relationships between panel members and the manufacturers to be pervasive rather than occasional, with panelists frequently being employees, consultants, or otherwise compensated by the company whose ingredient they were evaluating. [ESTABLISHED that conflicts are widespread; specific percentages vary by analysis — verify before quoting a number.]
Be precise about what that does and doesn't prove. It does not establish that these panels reach wrong conclusions, and plenty of them are staffed by genuinely qualified toxicologists doing careful work. Scientists with industry ties are not automatically corrupt, and it's worth saying that plainly.
What it does establish is that "generally recognized as safe" is doing rhetorical work its structure doesn't support. A determination made by a panel the manufacturer chose and paid is not a general recognition by the scientific community. It's a recognition by a selected subset of it, and the selection was made by the party with the financial interest in the answer. [CONTESTED as to consequences; ESTABLISHED as to structure.]
The case study: how something GRAS turns out not to be
The abstract version of this is easy to shrug off. The concrete version is the one that changed how we read labels.
Partially hydrogenated oils — the source of artificial trans fats — were treated as GRAS for decades. They were in margarine, shortening, packaged baked goods, fried food, and a huge share of the processed food supply. They arrived under the grandfather logic of the 1958 carve-out and stayed there through the entire second half of the twentieth century. [ESTABLISHED]
Then the science accumulated. Artificial trans fats turned out to have a genuinely bad cardiovascular profile — raising LDL cholesterol while lowering HDL, with an association to heart disease strong enough that essentially no one now defends them. In 2015, the FDA formally determined that partially hydrogenated oils are not generally recognized as safe, and they were removed from the food supply. [ESTABLISHED]
That's the system self-correcting, and credit where it's due — it did correct. But notice how long the correction took, and notice what made it possible: PHOs were ubiquitous, heavily studied, and consumed by nearly everyone, which is exactly the profile that eventually generates enough evidence to force a reversal.
Now apply that to a novel additive used in a handful of products, consumed by a small fraction of people, never notified to the FDA, and evaluated once by a panel the manufacturer hired. Where would the countervailing evidence come from? Who would generate it? That's the honest worry — not that GRAS substances are secretly poisonous, but that the system's error-correction depends on a scale of use and study that most individual additives will never reach. [CONTESTED — this is an argument about structure, not a demonstrated harm.]
Two other names from our library worth knowing here: BHA and BHT, the synthetic antioxidants, both entered American food in the 1940s and 50s under the pre-modern regime and were folded into GRAS status. Decades of subsequent research produced a genuinely mixed record on one of them — the scrutiny arrived long after the presence did. That inversion, evidence chasing adoption rather than preceding it, is the signature of this pathway. [ESTABLISHED]
The honest counter-case
We'd be building the haunted house we keep promising not to build if we stopped there. Here is the serious argument that GRAS is not the scandal this article has been implying, and it deserves a fair hearing.
Most GRAS substances genuinely are safe, and obviously so. The list includes vinegar, salt, sugar, baking soda, spices, common vitamins, and hundreds of other things whose safety is not in any real doubt. Judging the whole category by its most exotic members is the same error as judging all food additives by the scariest one. [ESTABLISHED]
Full pre-market review for everything would be enormously expensive and slow, and the cost would land somewhere. A system requiring a complete dossier for every substance would consume agency resources that are, in practice, finite — and it's a legitimate question whether spending them on cinnamon is a better use than spending them on the genuinely novel compounds. Regulatory attention is a scarce resource and triage isn't automatically corrupt. [CONTESTED]
Manufacturers face real liability. A company that puts a harmful substance in food faces lawsuits, recalls, and brand destruction. That's not nothing as an incentive, and it operates continuously rather than at a single approval moment. [CONTESTED]
The FDA retains post-market authority. GRAS status isn't permanent immunity — as the trans fat case shows, the agency can and does revoke it when evidence accumulates. [ESTABLISHED]
And the disaster record is not what the rhetoric implies. Despite decades of this system operating, there is no long list of GRAS ingredients that turned out to be acutely dangerous. That's genuine evidence about how the system performs in practice, and any honest critic has to account for it. [ESTABLISHED]
The strongest version of the criticism, then, isn't "GRAS is poisoning America." It's narrower and harder to dismiss: the pathway was designed for substances with obvious safety records and is now also used for novel compounds; the party with the financial interest chooses the reviewers; notification is optional; and consequently nobody — including the regulator — can say with confidence what's actually in the food supply. You can believe all of that and still believe most GRAS ingredients are fine. Both are true at once. [CONTESTED]
What this actually changes for you
Here's how to put it to work, because a concept you can't use is trivia.
Recalibrate what "FDA approved" means on a label or in an ad. For many additives, no approval occurred. If a marketing claim leans on FDA clearance, it's worth knowing whether that means a review, a "no questions" letter, or nothing at all. The phrase does much less work than it appears to. [ESTABLISHED]
Treat "it's been used for decades" as weaker evidence than it sounds. Long use is genuinely meaningful for substances humans have eaten for centuries. It's much weaker for a compound that's been in the food supply since 1994 and never been studied for chronic low-dose effects — the "decades" there mean it was adopted, not that it was examined. [ESTABLISHED]
Notice which ingredients are novel. The GRAS concern isn't evenly distributed. Salt is not the issue. The pathway's weak point is the modern synthetic compound with a short history and a small literature. When you see an unfamiliar chemical name, the useful question isn't "is this poison" — it's "how long has this been in food, and who looked at it?"
And apply the question that runs through our whole library. Almost every additive that raises real questions is a color, a texture agent, a preservative, or a shelf-life extender — something serving the product's appearance or economics rather than your nutrition. So the practical move is the same one it always is: does this thing do anything I actually need, and does a version exist without it? You don't need to resolve a regulatory debate to answer that in a grocery aisle.
The bottom line
GRAS began as a sensible exemption so nobody would have to file a federal safety dossier for vinegar. It became the pathway through which a large and uncounted number of substances — including novel synthetic compounds — enter the American food supply on the strength of determinations made by panels the manufacturers selected and paid, with notification to the FDA remaining voluntary and the agency's most common response being a letter that explicitly declines to approve anything.
Most of what came through that door is probably fine. Some of it, like partially hydrogenated oils, was not, and took decades to catch. And the structural fact underneath both statements is the one worth carrying with you: for a great many ingredients, the answer to "who decided this was safe?" is "the company that wanted to sell it."
That's not a reason to panic in the cereal aisle. It's a reason to stop outsourcing the question. The label tells you what's in there. This file tells you how it got there. What you do with both is, as always, yours.
We don't tell you it's poison. We don't tell you it's fine. We hand you the file, and you decide with your eyes open.
The full investigation — how GRAS works, the notification process, the conflict-of-interest research, and the specific ingredients that came through this pathway — is in The GRAS Files: How Companies Self-Approve Ingredients, available at alsofoundin.com/ebooks.
Browse the free ingredient library at alsofoundin.com/ingredients, or scan any label with the AlsoFoundIn app — seven free scans a day.
Approved ≠ Safe. Not Approved ≠ Unsafe.