I eat one meal a day, cooked fresh, whole food, sourced from places I trust. All of it deliberate, none of it on autopilot. And before any of it reaches my kitchen, there are ten seconds I never skip. I scan it. The app is Yuka. I point my phone at a barcode, and before I put something in my cart, a French app tells me what the label is trying not to. I use it on food and on the creams I put on my skin. It’s a small habit, and what makes it worth trusting is who the score answers to. Yuka is simple on purpose. You scan a product, and it returns a number from 0 to 100 with a color: dark green for excellent down to red for bad. For food, that number comes from three inputs. Roughly 60% is nutritional quality, the sugar, salt, saturated fat, calories, protein, and fiber that a Nutri-Score-style calculation weighs. About 30% is additives, ranked by risk against the published record. The last 10% is a small bonus for organic certification. It scores what’s in a product, not how much of it you’ll actually eat, which is a limit I’ll come back to. Here’s a Clif bar I scanned. Fifty-four out of a hundred, a pale green “good.” The breakdown is the part I like: eleven grams of protein and six of fiber on the positive side, two hundred and fifty calories and seventeen grams of sugar on the negative, no risky additives. Macro breakdowns like this are having a moment; Cal AI, which reads calories and macros off a photo of your plate, has been a fixture near the top of the health-app charts. But macros only tell you how much you’re eating. Yuka is after what, additives and processing included. It doesn’t tell me not to eat the bar; it ranks what’s in it, in the time it takes to reach for the shelf. My scan history is only half a grocery list. The things I actually buy come back excellent: coconut water, extra virgin olive oil, pasture-raised eggs. The Cheetos, the flavored beef jerky, the protein bar dressed up as health food come back bad or poor, and I don’t eat any of those. I scan them out of curiosity, because a red rating on something I’ve already written off is a cheap, standing reminder of why I did. The surprises are never the obvious junk. Everyone knows the Cheetos are junk. I didn’t know the “healthy” jerky scored worse than it looked. Scanning food matters because most of what’s for sale isn’t really food in the sense my parents would recognize. In the United States, [[ultra-processed foods::Ultra-processed foods: industrial formulations made largely from substances extracted from foods plus additives, with little or no intact whole food left. Category four of the NOVA classification.]] make up around 55% of the calories people eat. In British teenagers the figure is closer to two-thirds. These aren’t cakes and candy. They’re the breads, sauces, yogurts, and cereals engineered from ingredients you’d never keep in a kitchen. The evidence against eating this way has gotten hard to wave off. A 2024 umbrella review in the BMJ pooled data on nearly ten million people and linked higher ultra-processed intake to worse outcomes across dozens of measures, with the most convincing associations for cardiovascular death, type 2 diabetes, and anxiety. Association isn’t proof, and the review says so. But one of the few actual randomized trials, run by Kevin Hall at the NIH, fed people matched diets and found they ate about 500 calories more per day on the ultra-processed one, on menus balanced for sugar, fat, fiber, and salt. Something about the processing itself, not just the macros, drives the overeating. The additives are their own story, and they’re where the regulation gap gets embarrassing. Processed meat has been classified by the WHO’s cancer agency as carcinogenic to humans since 2015. Titanium dioxide, a whitener, was pulled from food across the EU in 2022 after European regulators decided they couldn’t call it safe. It’s still legal in American food. Red Dye No. 3 was only banned by the FDA in 2025, decades after the evidence that prompted it. The label won’t tell you any of this, because the label is built to sell the product, not to inform you about it. Yuka doesn’t stop at the score. It suggests alternatives, usually a similar product that rates better. It’s how I found half the things I now buy on default: a worse protein bar swapped for a better one, a sugary drink swapped for a plain kombucha. The nudge is small each time; over a couple of years it moves your whole basket. Once a scan settles a category, I tend to lock onto that one source rather than re-decide it every trip to the store, which matters most when I’m travelling and facing a shelf of brands I’ve never seen. None of that nudging is worth anything if the referee is bought. Plenty of things will rate a product for you: the retailer that stocks it, the influencer paid to hold it, the brand’s own front-of-pack claim, which is marketing with a nutrition panel attached. Every one of them has a reason to want the answer to come out a certain way. Yuka’s whole design is the absence of that reason. Its money comes from the people who pay for the app, not from the companies it grades, and in 2024 that subscription was essentially all of its revenue. There’s no ad to sell, no brand to keep happy, no data being resold behind the scan. I can’t audit every internal claim, but I can watch how it behaves under pressure, and it behaves like something with nothing to protect. The clearest test came in court. When Yuka flagged the added nitrites in cured meat as a cancer risk, the French charcuterie federation sued it for denigration. Three lower courts ruled against Yuka in 2021. It won all three on appeal, the last in June 2023, with the courts treating the warnings as protected speech on public health, backed by more than a decade of research. An app that an entire industry can take to court, and that wins, is answering to the record rather than to the brands. That’s the part I’m paying for. It also happens to be French, which I’ll admit I enjoy. It started at a Paris hackathon in 2016, first imagined as a scanner you’d stick on your fridge, until the founders realized the phone in everyone’s pocket already was one. It reached the US in 2020 and now counts tens of millions of users across more than a dozen countries. A quiet French export that grew by refusing to sell the one thing that made it worth using. I use Yuka on cosmetics more than people expect, because that market is even less honest than food. I scan moisturizers, face creams, and body lotion, and a couple of the brands I’ve since explored are ones the app pointed me to. The label problem is worse here: it tells you almost nothing you can act on. The regulation gap is not subtle. The European Union prohibits well over a thousand substances in cosmetics. The United States, under the law that governed the industry for most of a century, restricted around eleven. A 2022 reform called MoCRA gave the FDA real authority for the first time since 1938, but it still doesn’t approve cosmetics before they’re sold, and words like “clean,” “natural,” and “hypoallergenic” mean nothing in particular, because no agency defines them. Sunscreen is where this instinct goes sideways. Scan a chemical sunscreen and Yuka will often dock it for a filter like oxybenzone. There’s a loud movement online, mostly on TikTok, that takes the same ingredient and runs it somewhere else entirely: that chemical sunscreens are poisoning you and mineral filters are the only safe option. The kernel of truth is real. In 2019 and 2020, FDA studies showed that chemical UV filters get absorbed into the bloodstream above the level that triggers further safety review. That finding got flattened into “sunscreen is toxic,” which is not what it says. Absorption is not harm. The same researchers were explicit that no one should stop using sunscreen, and dermatologists have been consistent: use a broad-spectrum SPF 30 or higher, mineral or chemical, because the thing that actually causes cancer is the sun. A 2026 study of a thousand sunscreen videos found that most promoted sunscreen use, but the ones spreading misinformation drew the most engagement. The panic travels faster than the correction. Which is the honest limit of a tool like this, in cosmetics and in food alike: it scores the presence of a flagged ingredient, not the dose. A trace and a heavy load can land the same penalty, and for cosmetics the app doesn’t weigh concentration at all. Chemists have made this criticism fairly. A red rating means a product contains something with a question mark on it, not that it will hurt you. Hazard is not risk, and the app’s own simplicity makes the two easy to confuse. So I don’t read the number as a verdict. I read it as a prompt to look. Most of the time it confirms what I already suspected and I move on. Occasionally it flags something I’d have bought without a thought, and I read the ingredient list I’d otherwise have ignored. That ten seconds of attention is the whole return, and it pushes back against the default, which is to consume without noticing. That default is the part I care about most. What I’m protecting isn’t a biomarker or a score, it’s the habit of choosing, of not feeding myself by reflex. The body is the instrument the mind runs on, and the sharper you keep it, the better the work it returns. I decided long ago to be deliberate about the one meal I eat. Scanning the barcode is the same instinct pointed at the aisle: a small win that keeps me conscious about what I put in, with a second opinion that no one paid to give me.
