California’s Assembly Bill 2244 creates a state-sanctioned “Non-Ultraprocessed Certified” seal. Coverage of it has explained the standard by reciting California’s existing statutory definition of ultraprocessed food: a product is a UPF if it contains a listed additive and is high in saturated fat, sodium or added sugar (or contains a nonnutritive sweetener). FoodNavigator’s 1 September account sets out both limbs in exactly that way.
That is a correct description of Health and Safety Code § 104661. It is not the test AB 2244 uses to decide who gets the seal.
Section 104672(b)(1) of the enrolled bill bars certification of a food product that “[m]eets the additive criteria for ultraprocessed food, as described in subparagraph (A) or (B) of paragraph (2) of subdivision (a) of Section 104661.” It cites paragraph (2) — the additive limb. It does not cite paragraph (3), where the saturated fat, sodium and added-sugar thresholds live.
For a formulator, that inversion is the whole story. A 400-calorie product deriving 30% of its energy from saturated fat is certifiable if its ingredient list is clean of the eight named additive classes. A zero-sugar, low-sodium plant-protein cutlet held together with methylcellulose is not.
What the additive test actually is
Health and Safety Code § 104661 was added by AB 1264 (Stats. 2025, Ch. 467) and took effect on 1 January 2026. Its paragraph (2)(A) captures substances in the FDA’s Substances Added to Food inventory that are “designated as having any of the following FDA-defined technical effects”, and then lists eight, each by its citation in 21 CFR 170.3(o):
| # | Technical-effect class | 21 CFR 170.3(o) |
|---|---|---|
| 1 | Surface-active agents | (o)(29) |
| 2 | Stabilizers and thickeners | (o)(28) |
| 3 | Propellants, aerating agents, and gases | (o)(25) |
| 4 | Colors and coloring adjuncts | (o)(4) |
| 5 | Emulsifiers and emulsifier salts | (o)(8) |
| 6 | Flavoring agents and adjuvants (excluding 21 CFR 182.10 spices and natural seasonings) | (o)(12) |
| 7 | Flavor enhancers (same exclusion) | (o)(11) |
| 8 | Nonnutritive sweeteners | (o)(19) |
Section 170.3(o) defines 32 technical-effect classes. Twenty-four are not cited. The omissions are as consequential as the inclusions. Texturizers — (o)(32), “[s]ubstances which affect the appearance or feel of the food” — are absent. So are formulation aids (o)(14), which expressly cover “carriers, binders, fillers, plasticizers, film-formers”; enzymes (o)(9); firming agents (o)(10); humectants (o)(16); antioxidants (o)(3); antimicrobial agents (o)(2); and pH control agents (o)(23).
Paragraph (2)(B) then carves three things back out: salt or sodium chloride; spices and other natural seasonings and flavorings as listed in 21 CFR 182.10; and “[n]atural color additives, as listed in Part 73 of Title 21”. Those “shall not by themselves cause a food or beverage to be categorized as a UPF”.
Practical consequence: the certifiable/not-certifiable line is not drawn around processing, nutrition or ingredient origin. It is drawn around which of 32 boxes the FDA ticked for a substance in a database built from a 1972 National Academy of Sciences survey. That is the document a reformulation team should be reading, not the nutrition panel.
The database decides, and it does not use your label names
The statute keys off substance records, not ingredient declarations. Querying the FDA inventory on 13 September 2026 shows how far apart those two things sit.
| Substance (as named in the FDA inventory) | CAS / ID | Designated technical effects | Disqualifying classes hit |
|---|---|---|---|
| CELLULOSE, METHYL | 9004-67-5 | Emulsifier or emulsifier salt; flavor enhancer; stabilizer or thickener | 3 |
| HYDROXYPROPYL METHYLCELLULOSE | 9004-65-3 | Anticaking or free-flow agent; drying agent; emulsifier or emulsifier salt; humectant; processing aid; stabilizer or thickener; surface-finishing agent; texturizer | 2 |
| CARBOXYMETHYL CELLULOSE, SODIUM SALT | 9004-32-4 | Anticaking or free-flow agent; drying agent; emulsifier or emulsifier salt; formulation aid; humectant; stabilizer or thickener; texturizer | 2 |
| LECITHIN | 8002-43-5 | Anticaking or free-flow agent; antioxidant; drying agent; emulsifier or emulsifier salt; humectant; lubricant or release agent; masticatory substance; nutrient supplement; surface-active agent; texturizer | 2 |
| LECITHIN (VEGETABLE) | 977092-24-2 | Antioxidant; drying agent | 0 |
| BAKERS YEAST EXTRACT | 8013-01-2 | Flavor enhancer; flavoring agent or adjuvant; malting or fermenting aid; nutrient supplement | 2 |
| YEAST EXTRACT AUTOLYZED | 977082-78-2 | (no technical effect designated in the record) | 0 |
| BAKER’S YEAST GLYCAN | 977014-12-2 | Emulsifier or emulsifier salt; stabilizer or thickener; texturizer | 2 |
| BAKER’S YEAST PROTEIN | 977014-13-3 | Leavening agent | 0 |
| PHAFFIA YEAST | 977165-85-7 | Color or coloring adjunct (21 CFR 73.355) | 1, but carved out by § 104661(a)(2)(B)(iii) |
Read the second and fifth rows together. “LECITHIN” carries both emulsifier and surface-active agent designations and therefore disqualifies. “LECITHIN (VEGETABLE)” is designated only as an antioxidant and drying agent and therefore does not. A label that declares “sunflower lecithin” does not tell a certification agent which record applies.
The yeast rows are stranger still. “BAKERS YEAST EXTRACT” is a flavor enhancer and a flavoring agent, so it disqualifies. “YEAST EXTRACT AUTOLYZED” — the phrasing closer to what appears on many savoury labels — has an empty technical-effect field in the inventory, so on the statute’s plain wording it triggers nothing. “BAKER’S YEAST PROTEIN” is designated only as a leavening agent, which means a yeast-protein ingredient of the kind we examined in our piece on stated protein basis clears the additive test outright.
Practical consequence: certification will turn on record selection, and the bill gives no rule for it. Two manufacturers using the same commercial ingredient can reach opposite outcomes depending on which inventory row their certification agent maps it to. Anyone planning to seek the seal should document that mapping now: § 104675(c) requires the agent to make its certification records available to the department for inspection, and § 104675(d) lets the department audit those procedures and records “at any time”.
Where this bites the alternative-protein shelf
The technical-effect classes AB 2244 excludes are, with unhelpful precision, the ones that make structured plant and fermentation-derived products work. Binding and water management in meat analogues runs on methylcellulose and its relatives — all three cellulose ethers above disqualify. Emulsion stability in dairy analogues runs on lecithin and on gums designated as stabilizers or thickeners. Savoury character in fermentation-derived formats frequently comes from yeast extracts designated as flavor enhancers.
The colour carve-out is the one genuine piece of relief, and it is narrow. Part 73 of Title 21 lists colour additives exempt from certification — the “natural” colours. We looked at why a precision-fermented colourant that cannot reach Part 73 loses to beetroot red in our analysis of the colour route to market. AB 2244 adds a second, independent reason to prefer the Part 73 entry: a Part 73 colour is expressly carved out of the additive test, while a colour cleared by any other route is a “color or coloring adjunct” and disqualifies.
There is a structural point underneath. A conventional whole-muscle product needs no emulsifier, no stabiliser and no flavour enhancer, because the animal supplied the structure. An analogue has to build that structure from ingredients, and the ingredients that build it are the ones on the list. A seal designed around additive classes will therefore separate the two categories more reliably than a seal designed around nutrition would — which is the opposite of what much of the plant-based sector’s own nutritional case has argued for.
Three disqualifiers, on three different clocks
Section 104672(b) has three limbs, and only the first is self-executing.
| Limb | Test | What has to happen first |
|---|---|---|
| (b)(1) | Meets the additive criteria of § 104661(a)(2) | Nothing. The criteria are already in statute and took effect 1 January 2026. |
| (b)(2) | Is an “ultraprocessed food of concern” per § 104662 | CDPH must adopt defining regulations on or before 1 June 2028. Requires the full UPF test — additive limb and nutrient or sweetener limb — plus a “of concern” determination. |
| (b)(3) | Is a “restricted school food” per § 104662 | Same regulations, same deadline. |
The department must accredit certification agents “[n]o later than June 1, 2029”. So the regulatory definitions land roughly a year before the first agent can be accredited, and the additive test — the broadest of the three — governs from the outset without any rulemaking at all.
Two operating requirements sit alongside. Products must be recertified at least every three years (§ 104672(c)), and any reformulation requires recertification before the seal can be used on the reformulated product (§ 104672(d)). For a category that reformulates on ingredient-cost cycles rather than three-year cycles, that second clause is the expensive one.
The retail-placement requirement is smaller than reported
Coverage has described a duty on large grocers to display certified products prominently. The enacted wording is narrower. Section 114379.80(b) requires a grocery store with gross annual store sales above $10,000,000 to make “clearly identifiable” at least 3 or more certified items — but only “if the grocery store offers for sale more than 25 certified items”. Fuel retailers with a limited grocery line and convenience stores are excluded by § 114379.80(c). Section 114379.82(b) deems a store compliant if it “has made a good faith effort to comply”, and (c) excuses “[d]e minimis or isolated instances of noncompliance”. The whole chapter is repealed on 1 January 2040.
A store carrying 26 certified SKUs must make three of them identifiable. A store carrying 24 has no obligation at all. This is a threshold that a category manager can simply decline to cross.
Practical consequence: treat the retail clause as a merchandising opportunity rather than a distribution guarantee. The obligation it creates on the retailer is satisfiable with signage on three products.
The counter-argument
The reading above assumes the additive limb operates on its own for certification purposes. A fair objection is that this over-reads a cross-reference. Section 104672(b)(1) points to “subparagraph (A) or (B)” of paragraph (2), and subparagraph (B) is not a criterion at all — it is the carve-out saying salt, 182.10 spices and Part 73 colours “shall not by themselves” cause UPF categorisation. A product cannot sensibly “meet” an exemption. The coherent reading is that (A) states the test and (B) qualifies it, which is how this piece has treated it; but the drafting is loose enough that a court or the department could read the reference differently.
A second objection is practical: none of this binds anyone. The seal is voluntary. A manufacturer that would fail the additive test simply does not apply, sells the product as before, and loses nothing but a marketing claim. That is true, and it is why the retail-placement clause matters more than its narrow drafting suggests — it is the only part of the bill that attaches a consequence to not being certified. On current wording, that consequence is that a competitor’s three products get signage.
A third: the additive-class approach is not California’s invention. It borrows the FDA’s own functional taxonomy, which is auditable and already populated. A nutrient-based test would have needed new thresholds and new arguments; this one runs off an existing federal database. The cost of that convenience is the record-mapping problem in the table above.
What we could not establish
- Whether the Governor has signed. The Legislature’s own status page, read on 13 September 2026, shows “House Location: Governor” and a last action of “Enrolled and presented to the Governor at 3:30 p.m.” on 8 September 2026. Several outlets have described the bill as “approved” or “passed”; those refer to the legislative votes. We found no record of gubernatorial action and have not assumed one.
- The Senate vote count. The Legislature’s status page records the Senate third reading on 27 August as “Ayes 31. Noes 0.” FoodNavigator reports 32-0. Both agree the vote was unanimous and both give the Assembly concurrence as 72-0. We have used the official history for the Senate figure and cannot account for the one-vote difference.
- How certification agents will map ingredients to inventory records. The bill delegates certification procedures to the department (§ 104675(a)) and none exist yet. The lecithin and yeast-extract divergences above are our reading of the statute against the inventory as it stands, not a description of any adopted practice.
- Whether the “(A) or (B)” cross-reference will be read as we have read it. See the counter-argument above.
- Fee levels. Section 104677 authorises registration and certification fees “not exceeding the reasonable costs” of administration. No figure is set, and none is derivable from the text.
- How many currently marketed products would pass. Answering that requires ingredient-level data across a retail set, which we do not hold. We have not estimated it.
What to watch
- Gubernatorial action by 30 September 2026. The bill was presented on 8 September; FoodNavigator reports that Governor Newsom has until 30 September to sign or veto it.
- The CDPH rulemaking due on or before 1 June 2028 defining “ultraprocessed food of concern” and “restricted school foods”. Section 104662(b)(1)(G) instructs the department to consider “[w]hether the substance is a common natural additive” — the hook on which a fermentation-derived stabiliser or colour could be treated differently from a synthetic one.
- Whether the department publishes a record-mapping rule with the certification procedures. Without one, the lecithin problem is unresolved.
- Whether any alternative-protein manufacturer reformulates to clear the additive test rather than forgo the seal. Substituting a texturizer-only or formulation-aid-only ingredient for a designated stabiliser is technically possible and commercially unproven. The first company to publish a certified whole-cut analogue will have answered the most interesting question in this bill.