Method

Food Additive Index exists to answer one question well: where is a food additive legal, and where is it not. Everything else on a page supports that answer or cites it.

What we do not do

We do not rate additives. There is no score, no traffic light, no “safe” or “avoid”. Sites that lead with a rating are answering a different question, and a rating would misrepresent the underlying data, which is regulatory rather than evaluative.

We do not give health or dietary advice, and we do not resolve scientific disagreement. Where two bodies read the same evidence differently, both readings appear with their attribution and the site draws no conclusion of its own.

“Not listed” is not “prohibited”

This is the distinction the whole project turns on, and the one most often got wrong. A substance that is prohibited has been explicitly forbidden by a named instrument — 21 CFR 189, say. A substance that is not listed simply does not appear on that jurisdiction's permitted list. That is a statement about the list, not a finding against the substance.

The two are separate values in our database, they are rendered differently, and the constraint is enforced at the database level so that no ingest job can quietly collapse one into the other. Every jurisdiction appears on every additive page, including those with no entry at all, because omitting them would invite exactly the wrong inference.

Where the data comes from

We read regulators directly rather than secondary compilations. Status comes from each jurisdiction's own instrument; hazard reference values from EFSA's OpenFoodTox and WHO's database of JECFA evaluations; chemical identity from PubChem. The sources, with their licences and retrieval dates, are:

• Australia New Zealand Food Standards Code, Schedule 15 (compilation No. 16) — Food Standards Australia New Zealand; Federal Register of Legislation. CC BY 4.0; retrieved 2026-10-17.
• Health Canada Lists of Permitted Food Additives — Health Canada. Open Government Licence – Canada; retrieved 2026-10-17.
• Electronic Code of Federal Regulations, Title 21 — U.S. Government Publishing Office / National Archives. Public domain; retrieved 2026-10-17.
• EU food additives (Annex II to Regulation (EC) No 1333/2008) — European Commission. CC BY 4.0; retrieved 2026-10-18.
• EU additive names in 23 official languages (consolidated 02008R1333-20260818 and 02012R0231-20260818) — European Union (EUR-Lex). CC BY 4.0; retrieved 2026-09-19.
• EU food additive specifications and Annex III (consolidated 02012R0231-20260818 and 02008R1333-20260818) — European Union (EUR-Lex). CC BY 4.0; retrieved 2026-10-18.
• Great Britain food additive authorisations (FSA/FSS register; assimilated Regulation (EC) No 1333/2008, Annex II as at 28 June 2024) — Food Standards Agency / Food Standards Scotland; The National Archives (legislation.gov.uk). Open Government Licence v3.0; retrieved 2026-10-18.
• GRAS Notice Inventory — U.S. Food and Drug Administration. Public domain; retrieved 2026-10-18.
• Codex General Standard for Food Additives (CXS 192-1995) — Codex Alimentarius Commission (FAO/WHO). FAO terms — non-commercial use with attribution; retrieved 2026-09-07.
• Evaluations of the Joint FAO/WHO Expert Committee on Food Additives (JECFA) — World Health Organization. WHO website terms of use — © WHO, all rights reserved; non-commercial use with acknowledgement; retrieved 2026-09-20.
• Combined Compendium of Food Additive Specifications (JECFA), online edition — Food and Agriculture Organization of the United Nations. FAO terms of use — non-commercial use with acknowledgement of FAO and WHO; retrieved 2026-09-23.
• Japan: List of Designated Additives (Appended Table 1, Regulations for Enforcement of the Food Sanitation Act) — Digital Agency, e-Gov Law Search (law text); Ministry of Justice, Japanese Law Translation (English names); Ministry of Health, Labour and Welfare, 法令等データベース (standards for use). Public Data License 1.0 (compatible with CC BY 4.0) on all three sites — e-Gov, Japanese Law Translation and the MHLW website; statute text is excluded from copyright (Copyright Act, Art. 13); retrieved 2026-10-18.
• South Korea: Standards and Specifications for Food Additives (식품첨가물의 기준 및 규격), MFDS Notification, Part III — Ministry of Food and Drug Safety (식품의약품안전처), through the National Law Information Center (국가법령정보센터, law.go.kr). Not protected by copyright: Article 7 of the Copyright Act of Korea excludes notices and public notifications issued by the national government. The National Law Information Center states that its legal information may be used freely, commercial use included, under the Act on the Provision and Use of Public Data, and that it has no legal effect: the Official Gazette (관보) prevails; retrieved 2026-10-18.
• EFSA OpenFoodTox (2026-04-30) — European Food Safety Authority. CC BY-ND 4.0; retrieved 2026-09-20.
• Pages added by hand — Food Additive Index. Not applicable: a list of decisions, not data; retrieved 2026-10-18.
• PubChem (PUG-REST) — National Center for Biotechnology Information, U.S. National Library of Medicine. Public domain; retrieved 2026-09-29.
• Substances Added to Food (formerly EAFUS) — U.S. Food and Drug Administration. Public domain; retrieved 2026-10-18.
• Electronic Code of Federal Regulations, Title 21: sections read by hand — U.S. Government Publishing Office / National Archives. Public domain; retrieved 2026-10-18.

That covers 8 jurisdictions: Australia and New Zealand, Canada, European Union, Great Britain, Japan, South Korea, United States and Codex Alimentarius. Every fact row in the database carries a foreign key to the source it came from, with a resolvable URL and the date we retrieved it. If a fact cannot be cited, it does not ship. There are currently 65,499 regulatory provisions across 1189additives on this basis, last retrieved 2026-10-18.

Great Britain is not the United Kingdom here. Northern Ireland still applies the EU's additive rules under the Windsor Framework, so its status is the EU row. The Australia and New Zealand row is one jurisdiction because the Food Standards Code is one instrument in both countries, with the exceptions noted where they apply.

How the data is built

Each source is ingested by a separate job that archives the raw payload before parsing it, so a parse can be corrected without re-fetching a source that has since changed. Each job is idempotent: run twice against unchanged input, the second run changes nothing. Each run reports what it added, what it changed, what it dropped and why, and which near-matches it refused to merge automatically.

Deciding when a record from one source is the same substance as a record from another is the hardest part. We merge only on exact identifiers — CAS number, then E or INS number, then an exact normalised name. Anything less certain goes to a review queue for a person to decide. A wrongly merged additive would publish one substance's legal status under another's name, which is the worst failure this site could have.

The cost of that caution is duplicates: “BHT” in one part of the US code and “Butylated hydroxytoluene” in another arrive as two pages. A person decides when two pages are one substance, records the decision and its reason, and the pages are merged; the old address redirects to the page that remains. The same care applies to identifiers taken from enrichment sources. Where PubChem’s name search or EFSA’s export attached another substance’s identity to a page — its CAS number, its reference values — the match is excluded by hand and the values it brought are withdrawn.

What is written rather than imported

• Prose. Every additive has a plain-English summary and a description of what it is made from, with dietary flags drawn from that description. They are not imported from a source: they are drafted with an AI model, then checked against the status, identifiers and functions the page holds, and held to rules (no verdicts, no health advice, no ranking). How this site is made says what the AI does and does not do, and what has and has not been checked.
• Two committees, side by side. EFSA advises the European Union; JECFA advises Codex and, through it, much of the rest of the world. Both have set an acceptable daily intake for 99 additives here and differ on 35. A lower figure is not a finding that the other committee was wrong — an intake is a reference value, not a threshold above which harm begins, and several differences are a recent opinion beside a much older one. Both are shown with the year each was set, and this site chooses between them nowhere.
• Studies and interpretations. Curated, with the same AI-assisted drafting, for a small number of substances where regulators read the same evidence differently, and shown with both readings. EFSA’s re-evaluations of the additives authorised in the EU before 2009 are summarised on each page they cover, in EFSA’s own framing, with the opinion linked.
• How each regime works. Every additive page shows the same jurisdictions, including those with nothing to show. What an absence means differs completely between them, so each has a page of its own saying what the regulator reads, what this site imports, and what an empty row does and does not prove.
• Allergen origin. Additives made from, or that may be made from, a substance on the EU allergen list (Annex II to Regulation (EU) No 1169/2011) are marked, with the exceptions that Annex II names. This describes sources, not products: whether a food must declare an allergen depends on what its additive was actually made from.
• Labelling obligations. A few additives oblige words on the pack: six colours carry a statement about activity and attention in children, aspartame a declaration for people with phenylketonuria, more than a tenth polyol a laxative statement, and in the United States one colour must be named rather than called a colour. Each is quoted from the instrument that imposes it. That a regulator requires a sentence is a fact about the regulation; this site takes no view on what any additive does.
• Everyday food groups and starting materials. The groups that make levels comparable are mapped by this site rather than read from a source, and so is the starting material each additive is recorded against — a description of a route, not a risk and not a dietary answer.
• The vocabulary. Terms of art decide what a page means, and several mean close to the opposite of what they look like. They are defined once, in the glossary, and pages link to them rather than re-explaining them in a parenthesis.
• The documents themselves. Every citation assumes the reader knows what Annex II or 21 CFR 184 is. Instruments says what each document covers, how it is arranged, what this site takes from it and what it leaves.
• The change feed. Imports update statuses on their own, but deciding that a new instrument is a revocation, a stay or a technical correction means reading it. The 19 entries on Changes are written that way, each citing the instrument, and new documents from the Federal Register and EUR-Lex are listed for review rather than published automatically.

Functions, comparisons and changes

• Functions come from the regulators: Codex’s functional classes, the US technical effects of 21 CFR 170.3(o), Health Canada’s purposes of use, and the EU’s separate lists of colours and sweeteners. Grouping a US “Non-nutritive sweetener” with a Codex “Sweetener” is this site’s judgement of equivalence, so every function page shows each regulator’s own term beside it. Regulators do not rank an additive’s roles, so the line at the top of a page names the function or functions the most regulators assign, and says which.
• Comparisons never call a missing permission a ban. Each difference is sorted by what can be shown: an explicit prohibition, a removal, a related listing that may cover the substance, a substance regulated under a different law, or an absence this site cannot confirm (the US permits self-determined GRAS use without a listing; Canada and Japan are matched by name, Japan permits existing additives outside the list this site imports, and South Korea is matched by identifier but only partly imported).
• Detected changes are found by comparing each week’s import with the one before. They are dated by when they were seen, not when they took legal effect, and they are not interpreted. Each one names the provisions that moved — the food category, and the figure before and after — because a count of provisions added and removed tells a reader that something happened and nothing about what.
• Reference values are compared the same way. When EFSA or JECFA revises an acceptable daily intake, the change is recorded against the additive with both figures and the year each was set. It is never described as a tightening or a relaxation: the two committees work to different terms of reference and on their own timetables, and a lower figure is not a finding that the other was wrong. An intake is also not a legal limit — it binds nobody until a regulator adopts it.
• The EU re-evaluation programme has a page of its own. It is the Regulation — deadlines, the additives named in each priority list, and the seventeen re-evaluations already finished when the programme began — and not a scoreboard of EFSA’s progress, which this site does not import. An additive that says nothing there is one this site holds nothing for, not one whose re-evaluation did not happen.

How much, and what the label must say

• Quantity is four answers, not one. A listing may set a figure, or say quantum satis, or good manufacturing practice, or nothing. Only the first is a number, and the middle two are limits — the amount the technological effect requires and no more — so rendering them as an empty cell would repeat the mistake of reading an unlisted substance as a prohibited one. They are separate values here and rendered differently.
• Ceilings are compared only where that is legitimate. Two figures mean something against each other when the food categories describe the same foods, so the comparison is read through the site's own food groups, each figure is printed with the category it came from, and what is compared is the most permissive figure each system sets anywhere in the group. A listing stating several ceilings at once takes no part: its highest figure is not known, and Great Britain states compound limits far more often than the EU does, so including them would manufacture a divergence out of a drafting habit.
• Specifications are compared on figures only. Where the EU and JECFA both specify a substance, the parameters each puts a number on are set side by side. Only the contaminant ceilings are ranked: JECFA's assay for annatto is 25 % against the EU's 85 % because the two specify different preparations, not because JECFA would accept a quarter-strength colour.
• Carry-over explains most apparent contradictions. An additive can be in a food the permitted list does not cover because it was permitted in an ingredient and came along with it, or because it is a carrier inside another additive. Both are set out, because a permitted list read alone does not predict a label.

Known gaps

• Other jurisdictions. 8 today. Absence of a country from this site says nothing about its rules. India is the one actively wanted and not built: the Food Safety and Standards Authority of India serves the same four-kilobyte bot-challenge page for every path, including direct links to its own PDFs, and defeating that is not something this project does. The jurisdiction is deliberately not added to the database ahead of the data, because every additive page shows every jurisdiction and an empty India column would be 804 assertions this site cannot support. China is the other large absence: the two editions of GB 2760 served as text are both superseded, the current one is rendered by script from a portal that serves nothing to a plain request, and no term published with it permits republishing its content.
• INS numbers are missing from 431 pages. The international numbering system is assigned by a Codex guideline, CXG 36, and every FAO route to it refuses this site’s requests — the same control that puts the Codex food additive standard behind a manual download. Most of the 431 are US-only substances that may never have been given a number; 82 carry an E number and almost certainly have one. The site does not guess: an INS number is not derived from an E number, however alike the two systems look.
• Specifications are two documents, not one. The EU sets identity and purity by regulation; JECFA publishes its own specifications through FAO, which are advisory until a country adopts them. Where both exist they are shown separately, because they are different documents written by different bodies on their own cycles.
• US GRAS notices. FDA’s responses to GRAS notices are shown on a page only where the notice names the substance exactly as this site does, and they are shown apart from the permissions: a response concerns one notifier’s conclusion for one use, not a listing in 21 CFR. Most notices concern substances with no page here. The provisional listings of 21 CFR 82 (the FD&C lakes) are included and marked provisional.
• Japan is partial. The list of designated additives is imported, matched by name through the Ministry of Justice’s unofficial English translation, together with the standards for use that limit many of them. Those standards are shown in the Japanese they are written in, because no official English text exists; what is in English beside them describes what the rule does and is this site’s, not the Ministry’s. Japan’s existing additives, natural flavouring agents and ordinary foods used as additives are permitted outside the designated list, so a page with no Japan row says nothing about Japan’s rules.
• South Korea is partial, and checked harder. Part III of the Ministry of Food and Drug Safety’s Standards and Specifications for Food Additives is imported, with each entry’s main use and standard for use, in the Korean it is written in; what is in English beside a standard describes what the rule does and is this site’s, not the Ministry’s. The notification prints INS and CAS numbers, so entries are matched by those first and by exact name only after, and an entry whose identifiers disagree with a page’s (a fatty acid and the generic fatty-acids page, a salt and its parent) is held for review and not matched. Mixed preparations and sanitisers for utensils are not imported. A new version applies from its effective date and not its publication date: the amendment published in July 2026 does not take effect until 2028 and is not what this site shows.
• Matching by name leaves names behind. Health Canada's lists and Japan's designated additives carry no E or INS numbers, so a listing reaches a page only where the name identifies the substance exactly. The names that did not match arepublished rather than left silent, so an absence on a page can be told apart from a gap in this site. The 197 names standing there in September 2026 were read one at a time against each source’s own CAS numbers and synonyms: 30 were pages this site already had, reachable only through a synonym or a number written as a group, and the rest were substances with no page at all, which now have 166 pages between them. None is left waiting. Each decision is recorded with its reason in the import’s own file, and nothing is settled by loosening what counts as a match — a page created for a substance is how its listing gets recorded at all, since an additive with no page cannot hold a permission. Where a regulator’s name does not say which form or which source it means, the page says that too, rather than choosing one.
• Sub-forms. Codex numbers some forms separately — INS 322(i) and 322(ii) for lecithin — where the EU lists one E number. Where the EU specification names the form, the page says so; where it does not, the page points at the related listing and says it cannot tell.

Corrections

Every page names its sources and the date. If a page disagrees with the regulation it cites, the regulation is right and we are wrong — check the retrieval date first, since regulations change and ours may be stale.

2 October 2026. The “Detected in the weekly refresh” section ofChanges, and the same entries on additive pages and in the Atom feed, held fifteen differences dated 24 September, including Canada apparently dropping glycerin, tocopherols and trisodium phosphate. None of them was a change in any regulation. They were this site's own work on identifying substances, read by the comparison as though a regulator had acted: the Japanese, Australian and New Zealand and EU source documents were identical from one run to the next, every Great Britain provision cited an Annex II fixed at 28 June 2024, and the Canadian provisions came back, in the same numbers, when a matching fault was fixed on 29 September. All fifteen were removed. The comparison now reports only what a source changed.

What is proposed

Decisions that have been asked for but not made are collected on Proposed: applications on the British register, draft regulations in the Commission's comitology register, petitions the FDA has under review, applications before Food Standards Australia New Zealand. An application is not an authorisation and a proposal is not a prohibition; no outcome is predicted, and each entry links the register it came from so the current position can be checked there.

Reading a label

The label lookup takes a pasted ingredients list, or a photograph of one. Both are handled entirely in the reader's browser: the text recognition engine, its WebAssembly core and its English language data are served from this site and run on the reader's own device, so neither the picture nor the text it produces is uploaded, and no third party is sent any part of it. Recognition is imperfect, so the text it produces is put in the box for the reader to check rather than used silently.

Labels in the other official languages of the European Union are matched as well, from the names the Official Journal prints in each language — acide citrique,Citronensäure, κιτρικό οξύ. Those names are fetched only when something in a pasted list is not recognised, so the page stays small for a reader who does not need them.

What this site measures

Every page loads Google Analytics, which records that the page was viewed and roughly where from; it sets its cookies only if you choose “Accept all” in the box that asks on your first visit. It is there to answer one question — which of these pages anyone actually reads — and the answer decides what gets written next.

It does not see what you type. The label lookup matches a pasted ingredients list, or a photograph of one, entirely in your own browser, and neither the text nor the picture is sent to this site or to anyone else; that is a property of how the page is built, described above, and analytics does not change it. Google does see the address of the page you are on, which on this site is the name of an additive. Privacy sets out the whole of it, including the two other requests a page makes and how to block any of them.

Reuse

There is a stable JSON document for every additive under /api/v1/, described by a JSON Schema those documents point at, andbulk CSV files holding everything this site is licensed to redistribute. Both are given on request: ask for a key, say what the data is for, and a person decides. A dataset description is at /llms.txt. What the bulk files leave out, they leave out for licensing reasons that are stated on thedata page rather than silently. Please carry through the retrieval date when you quote a status: a regulatory status without a date is not a fact. Source licences differ and are listed on Sources — some of the underlying data is not ours to sublicense.