Key takeaways:
- The FDA and HHS sent OMB’s regulatory review office a white paper, not a proposed rule, defining ultra-processed food on August 3, 2026. Filed as a notice with no legal deadline, it creates no binding labeling or formulation requirement by itself.
- The Senate is pushing from the opposite direction with actual legislative teeth. The Childhood Diabetes Reduction Act (S.5026) cleared committee 12-10 on July 22, and the Food Labeling Modernization Act followed a week later. Neither uses the FDA’s trigger criteria.
- California already has a legal UPF definition in force, built on functional ingredients and nutrient thresholds that match neither FDA’s nor Congress’s approach. Three definitions are live at once, and they don’t agree with each other.
On August 3, 2026, FDA and HHS sent a document called “Proposed Definition of Ultra-Processed Food” to the Office of Information and Regulatory Affairs, the U.S. Government’s Office of Management and Budget’s (OMB) regulatory review desk. It carries a RIN, 0910-ZD60. It is not a proposed rule, an advance notice of rulemaking, or a guidance document. It’s filed as a notice, the category OMB uses for informational submissions with no legal deadline and no “economically significant” designation.
Publishing this paper, whenever OMB clears it, won’t put a new label on a package or take an ingredient off a formulation. It puts a federal definition on paper that anyone (e.g., a state legislature, a retailer, a plaintiff’s attorney) can point to. Whether it ever becomes enforceable is a separate, later decision the administration hasn’t made yet.
Why a white paper carries no legal force, yet
A proposed rule comes with a comment period, a target date, and a legal path to enforcement. A white paper doesn’t, so it won’t force labeling and formulation changes on food manufacturers. Getting from this paper to an enforceable standard, on the label, in federal procurement, or inside school nutrition programs, means a separate rulemaking for each one, with its own notice-and-comment process and timeline.
The FDA also filed this one alone. The USDA, which regulates meat, poultry, and school meals, isn’t a co-signer, and that’s a gap that has to close before any single national standard could apply across the food supply. HHS Secretary Robert F. Kennedy Jr. has said he’d rather get to reformulation through voluntary industry commitments than a lengthy rule. That’s a stated preference with no rulemaking or date attached to it yet.
Two Senate bills are writing their own, conflicting definitions
While the FDA was filing a notice, the Senate HELP Committee was voting on something with teeth. The Childhood Diabetes Reduction Act (S.5026), introduced by Senator Bernie Sanders, would require the FDA to put warning labels on the front of ultra-processed packages and ban junk food advertising aimed at kids. It passed committee 12-10 on July 22, 2026, according to the committee’s own release and the full bill text. The National Association of Manufacturers has already warned it could “unfairly target nutritious products,” and its odds on the floor are uncertain.
Look at what really triggers a warning under that bill, though: added sugar, saturated fat, sodium, or high-intensity sweeteners past a threshold, a nutrient test, not FDA’s ultra-processed category. A product could clear the FDA’s definition and still get a warning label under the bill, or the reverse.
A week later, Senators Blumenthal and Booker introduced the Food Labeling Modernization Act (companion to House bill H.R.8385), which takes the opposite approach. It writes FDA definitions for “ultraprocessed” and “not ultraprocessed” into a standardized front-of-package labeling system. Two bills, both aimed at the same shelf, running on two different definitions of the thing they’re both trying to label.
California’s AB 1264 is a third definition, and it’s already law
While Washington argues over whether ultra-processed needs a legal definition at all, California already wrote one into law. Governor Newsom signed AB 1264 on October 8, 2025. It defines a covered product as anything containing at least one listed functional ingredient (e.g., emulsifiers, non-natural colors, non-natural flavoring agents, flavor enhancers, stabilizers, or non-nutritive sweeteners) combined with a nutrient threshold: 10% or more of calories from saturated fat, 10% or more from added sugar, or sodium in milligrams equal to or exceeding the calorie count, according to ArentFox Schiff’s summary of the bill.
Right now it only reaches K-12 school meals and competitive foods, with a synthetic-dye ban starting December 31, 2027, school phase-out beginning July 1, 2029, and full exclusion from reimbursable meals by July 1, 2035. But it’s already law, built on its own functional-ingredient logic that lines up with neither the FDA’s forthcoming definition nor the Senate bill’s nutrient triggers. A product engineered to clear one of these three frameworks has no guarantee of clearing the other two.
What three competing definitions mean for your reformulation budget
The FDA definition isn’t the finish line, and it might not even ship as a rule. Building a multi-year reformulation plan around one specific NOVA-style definition, before the FDA has committed to making it binding, means betting capital on a document RFK Jr. himself has floated skipping past in favor of voluntary commitments.
Congress and California are both moving faster than the FDA, and either can bite first. S.5026 has a floor vote ahead of it and real industry opposition, so its odds are genuinely uncertain. California’s is already signed into law. The FDA’s white paper, by contrast, doesn’t even have a date attached to becoming a rule.
With three incompatible definitions in play, the near-term work is measurement. Whether a SKU trips a functional-ingredient threshold, a nutrient threshold, or a NOVA-style processing category are three different questions with three different answers. Most reformulation teams are still answering them by hand, SKU by SKU, across whatever ERP, LIMS, and spreadsheet holds the formulation data this quarter. That’s the same data-fragmentation problem showing up in AI adoption research across the sector: the tool matters less than whether the underlying ingredient and nutrient data sits in one place a system can actually query.
Nobody has finished writing the rule that will eventually apply to your labels. Three separate groups are drafting three separate answers at the same time. Whichever one lands first, the plants in the best position that day will be the ones that already knew where every SKU stood under all three.
