Why AI Can't Tell You Which Omega-3 Is Actually Good

I build omega-3 for a living, so one evening I did something slightly masochistic. I opened an AI search engine and asked it how to choose a good one, because I wanted to see exactly what my customers see. The answer came back in seconds, confident and tidy. And as I read it, my stomach sank, because most of it was wrong.

Why you shouldn’t trust AI about omega-3

The model wasn't careless. It repeated its sources faithfully. The trouble is who those sources are. The pages it leaned on were magazines, health blogs, and nutritionists: confident voices, and not one of them makes oil. I've spent years on the manufacturing side, and the facts that actually decide quality aren't written down anywhere for a search engine to find. The people who know them are usually too busy making the product to publish. So the internet fills the gap with people who write well instead of people who make well. That is the whole reason I started writing these notes.

I see the same thing land in our inbox every week. Since ChatGPT arrived, the questions look more expert than they used to: the right words, the right form names, the confident tone. I like that people are doing their homework. But a sentence or two in, I can usually feel that a machine assembled the question, not a person who understands the thing underneath it. The vocabulary is excellent. The understanding isn't there yet. That is not the customer's fault. It is exactly the gap I want to close.

Back to my little experiment. The advice came down to three things: pick the oil with the highest percentage of EPA and DHA, choose ethyl ester because it's cheaper and supposedly more stable, and look for a third-party heavy-metal test. Sensible, if you've only ever read about oil. Let me walk you through why each one let me down.

The percentage trap

Take the first tip, the high percentage. Natural omega-3 sits at a modest concentration. Fish oil is around 30 percent EPA and DHA because that is what fish actually produce. Algae, which is where fish get their omega-3 in the first place, sits in the same modest range.[3] To get far above that, you have to concentrate the oil in a factory. In the classic process, roughly 30 percent gets converted to ethyl esters and then molecularly distilled up to around 60 percent, stripping the oil apart and rebuilding it to hit a number. [1] There is no gentle way to reach a 60 or 70 percent oil.

So when I see a very high number on a label, I don't read quality. I read heavy processing. If you want the oil closest to the food it came from, you want honest numbers, not big ones. A lower concentration often indicates that someone left the oil alone. You can't out-engineer nature. You can only push the oil further from it.

Why "stable" ethyl ester isn't

The second tip is where the AI was half right. Ethyl ester is cheaper. That part is true. Stable and well absorbed, it is not. It's a concentrated, factory-made form where the fatty acids are hung on an alcohol backbone instead of the glycerol one they naturally use, and your gut has to do extra work to break that bond.

This isn't just my opinion from the factory floor. In the landmark head-to-head trial by Jørn Dyerberg, often called the father of fish oil research, the re-esterified triglyceride form put roughly 70 percent more EPA and DHA into people's blood than ethyl ester did. Measured against natural fish oil at 100 percent, the rebuilt triglyceride landed at 124 percent and ethyl ester came last at 73 percent. [1] Calling the cheap form the stable, sensible choice gets the trade-off backward. You save money on the day you make it, and the oil quietly pays the bill later.

Three forms, and which one you're really getting

People ask me which form is best, so here is the part no label explains. There are three forms on the market, and the difference between them is most of the game.

Here's the catch I wish more people knew. A lot of oils sold as "triglyceride form" are only partly the real thing. In one analysis, some products marketed as triglyceride were only about 55 to 60 percent genuinely rebuilt triglyceride.[2] It still absorbs well, but it isn't the original form the oil came in, and that distinction never makes it onto the box.

The dark glass and the testing badge

The third tip was packaging and testing. Half right again. Light does damage omega-3, so dark pharmaceutical glass is a sensible choice, and clear or plastic bottles aren't. But the heavy-metal badge has quietly turned into a box everyone ticks. Refined oil gets cleaned during processing, so a clean metals panel is close to a given. It tells you almost nothing about the two things I actually lose sleep over: the form of the oil and its freshness.

The certificate that isn't theirs

Which brings me to the certificate. This year, our whole industry discovered transparency, and suddenly everyone posts a lab result and a TOTOX number. The first time I saw competitors doing it, I thought, good, finally. Then I looked closer.

And the shelf is worse than most people assume. When researchers in New Zealand pulled fish oils straight off the shelf and tested them, only about 8 percent passed all three freshness limits. Eighty-three percent were over the peroxide line, and half were over TOTOX. [6] The bar they were failing isn't even strict: the industry's own voluntary limits are a peroxide value under 5, an anisidine value under 20, and a TOTOX under 26.[4]

So when a brand shows me a spotless number, my first question is whose oil it describes. Most of the time, those numbers aren't theirs. They're copied from the raw-material supplier's certificate, measured on the bulk oil in a drum before anyone touched it. The number is real. It just describes the oil at the wrong moment. Because that drum gets opened. The oil meets air, gets pumped, gets filled into capsules. Oxidation starts right there and keeps going on the shelf. A peroxide or anisidine value taken at the source tells you nothing about the oil someone swallows six months later.

So when a brand shows me a certificate, this is what I actually want to see:

• Stability data for the final product, the capsule or bottle, not the bulk oil.

• Oxidation measured over time, at least at six months, not just on day one.

• A certificate with the batch number, the manufacturing date, and the testing date on it, so I can see the gap between them.

If all a brand can show is the supplier's drum, they're handing you someone else's homework.

What no label can show you

And then there's the part nobody outside the factory can check at all: no search engine, no expert squinting at a label.

• Nitrogen protection. Oxygen is what turns a good oil rancid. We flush the capsule and bottle with nitrogen to push the oxygen out before it reaches the oil. It's one of the most important calls a maker makes, and it appears on no label.

• The true triglyceride share. Real rebuilt TG versus the cheap concentrate, invisible to the buyer.

• How it ages. Every oil looks fine on day one. What matters is how it holds up after a year, after two, and how fast it turns once you open the bottle and let the air in.

Our side of the label

So let me turn our own label around and show you the back of it.

It starts with the ingredient, not the machine. We pick a natural oil at an honest concentration and treat it gently, because freshness is won at the source, not rescued at the end. A cheap, over-processed oil is already losing before it reaches a bottle.

Then we guard it. We nitrogen-flush the fill so the oxygen is gone before the oil ever meets air. It costs us, and it's only one piece of the system, not the whole thing. The good starting oil, the gentle handling, and the nitrogen only work together. No single step saves an oil on its own. That's the standard behind our omega-3 range.

Then we test the finished product, not a launch-day sample. We run stability tests on our own capsules, not the supplier's drum, at 3, 6, and 12 months, and again at the end of shelf life. We measure peroxide value and anisidine value, the numbers that actually track oxidation. TOTOX gets quoted like a headline, but even researchers describe it as a convenient composite rather than a rigorous measure, so we publish the two values it's built from instead of hiding behind the single figure. [5] Every certificate carries the batch number, the make date, and the test date, so you can see the time between them.

And we run the test I care about most, the open-bottle test. The moment the seal breaks, air gets in and oxidation starts, so we measure how long the oil really lasts in use, not on paper. That's why we ask you to finish a bottle within two months of opening it. After that, I can't stand behind the freshness, and I'd rather sell you a smaller bottle you finish fresh than a big one that turns on your shelf. None of this fits on a label. None of it shows up in an AI answer. It only shows up in how the oil behaves a year from now.

The only honest advice I have

So here's the only honest advice I can give you, and it's uncomfortable coming from someone who sells the stuff. You can't verify the invisible things yourself. Standing where you stand, even I can't fully vouch for another brand's oil without their data. Trust isn't optional here. There's no clever way around it.

The real question isn't whether to trust a brand. It's how to choose one, and how to make it earn the trust. So ask. Ask a lot. A brand that made real choices will have real answers. One that didn't will change the subject. These are the questions I'd ask:

• What form is the oil, and how much of it is true triglyceride?

• Is it nitrogen-flushed in the capsule and bottle?

• What's the oxidation value of the finished product at six months, with the batch and the dates, not the supplier's drum figure?

• Where does the oil come from, and who tested it?

The internet has argued about absorption and bioavailability for years. Useful, but it skips the first question: was the oil kept alive long enough to absorb anything at all?

And the simplest check of all is the price. Real ingredients, gentle work, nitrogen, honest testing, all of it costs money. Quality has a floor, and it isn't low. A high price doesn't guarantee quality, but a low one rules it out. Behind every price there's a decision. Look at a cheap oil the way an economist would, and you start asking what got sacrificed to hit that number. That's why we'd rather answer the hard questions than dodge them.

Michal Koci, Founder