Imagine an agency that tells an entire industry, in final guidance published after a formal comment process, that a product may be labeled a certain way. Imagine a company that spends years and substantial resources building a product in reliance on that guidance—then goes further, negotiating its proposed label with the agency, accepting the agency’s revisions, and obtaining the agency’s approval, published in the Federal Register.

Now imagine the agency appearing in court six years later to announce that its guidance meant nothing and its approved label meant nothing—and a federal court of appeals holding that the company—which took the agency at its word—is fair game for state-law consumer-fraud class actions. Last week, in Franco v. Chobani, LLC, No. 25-2087 (7th Cir. July 27, 2026), that is precisely what the Seventh Circuit did.

Every regulatory compliance lawyer who has ever told a client “the agency has signed off on this” should read the opinion and shudder.

Six Years of Green Lights:

Allulose is a naturally occurring sweetener that the human body barely metabolizes. It contributes almost no calories, does not raise blood glucose or insulin, and does not cause cavities. In 2020, after receiving citizen petitions, publishing draft guidance, and taking public comment, FDA issued final guidance announcing that allulose need not be counted in the “Total Sugars” declaration on food labels. The agency explained that its thinking had moved beyond a rigid chemical-structure test toward the statute’s actual purpose: helping consumers maintain healthy diets. The guidance was published in the Federal Register. It was directed at the entire food industry. It said what it said.

Chobani took the government at its word. It spent more than two years developing and patenting a yogurt made with ultrafiltered nonfat milk and sweetened with allulose—Chobani Zero Sugar. And it did not stop there. Because ultrafiltered milk falls outside FDA’s official definition of yogurt, Chobani sought a temporary marketing permit—a process that, by regulation, required Chobani to submit its proposed label for approval to FDA’s Office of Nutrition and Food Labeling. The agency did not rubber-stamp it. FDA scrutinized the label, demanded revisions to the very portions addressing allulose and sugar content, blessed the revised label, and then published that approval in the Federal Register in March 2023.

A published draft. Public comments, received and addressed. Final guidance in the Federal Register. A label submitted to, revised by, and approved by the agency—also published in the Federal Register. If regulatory reliance means anything, it means this.

The About-Face:

Then came the lawsuit. The Francos, would-be class representatives, sued under Illinois consumer-fraud law, arguing that “zero sugar” is deceptive because allulose is, chemically, a monosaccharide—and FDA’s regulation defines sugars as “all free mono- and disaccharides (such as glucose, fructose, lactose, and sucrose).” The district court dismissed on preemption grounds, sensibly finding that FDA’s considered guidance controlled and that the agency could not simultaneously permit the label and invite juries in fifty states to punish it.

On appeal, the Seventh Circuit invited FDA to weigh in. And FDA—the same agency that wrote the guidance, ran the comment process, and approved the label—filed an amicus brief agreeing with the plaintiffs that “all” means all. Allulose is a sugar after all. The 2020 guidance? Merely a statement of enforcement discretion, not an interpretation of anything. The agency’s own considered, published position was recharacterized, six years later, as non-binding navel gazing.

The panel agreed. It held the regulation clear, dismissed the guidance as a mere enforcement-discretion musing unworthy of weight, and noted that under Kisor v. Wilkie (2019), deference would likely have been inappropriate anyway because the guidance conflicted with the agency’s earlier positions. But strip away the doctrinal labels and look at what happened: the agency’s eleventh-hour litigating position prevailed over its published, commented-upon guidance and its Federal Register-published permit.

Critics of Auer deference have long feared that agencies would write vague rules and then interpret them opportunistically. Franco shows that agencies have found a better trick: guidance informal enough to escape the rigors of rulemaking, and—when convenient—too informal to bind the agency to anything it said. And consider the irony: Kisor’s bar on inconsistent agency interpretations—a rule built to prevent flip-flops—was the court’s reason for discounting the guidance that Chobani relied on. The actual flip-flop, the government’s brief, sailed through. Heads the agency wins; tails the regulated party loses.

“A Sophisticated Actor”:

The court’s answer to Chobani’s reliance argument deserves quotation, because it proves far too much. Chobani, the panel wrote, “is a sophisticated actor” that should have known FDA’s enforcement decisions “would not immunize the company” from state-law suits.

Take that logic seriously and no FDA guidance document—and no temporary marketing permit—ever provides meaningful protection, because the plaintiffs’ bar and fifty state consumer-protection regimes always remain open for business. Yet guidance is how FDA actually regulates the food supply. The agency has issued hundreds of guidance documents precisely because rulemaking is slow and resource-intensive. Indeed, the 2020 guidance promised “future rulemaking” on the Total Sugars definition. Six years later, as of this writing, that rulemaking has not happened. Companies cannot wait for rules the agency never writes; they must act on the signals the agency actually sends.

The Supreme Court has recognized exactly this problem. In another case involving a sophisticated actor, Christopher v. SmithKline Beecham Corp. (2012), the Court refused to credit an agency interpretation announced for the first time in an amicus brief, because imposing massive retroactive liability on parties who relied on years of agency acquiescence would work an “unfair surprise.” And in Christopher, the industry had relied on nothing more than agency’s silence. Here, Chobani had the agency’s sign off, memorialized in the Federal Register. Fair notice is not a courtesy; it is a measure of due-process—yet Christopher appears nowhere in the panel’s opinion.

The Lesson:

This is not a story of whipsawing elections. The 2020 guidance issued under President Trump’s FDA. The marketing permit was reviewed and granted under President Biden’s. And the brief recharacterizing the guidance as non-binding musing came from the FDA of President Trump’s second term—the same president whose agency made the original promise. Three administrations, both parties, one consistent signal to industry. And when the Seventh Circuit asked the agency what its regulation meant, FDA was free to stand behind its guidance, or at least to urge the court to protect those who had relied on it. It did neither. Instead it disowned the guidance. That is not politics. It is something more corrosive.

Government regulators should not bait and switch. When an agency speaks to an entire industry through published guidance, and then confirms its position by negotiating and then approving a specific company’s product label, elementary fairness forbids the agency from appearing in court to disavow it all. Franco tells every regulated business in the Seventh Circuit that FDA guidance is not a shield but a trap, and that the only safe course is to demand formal rulemaking the agency rarely and begrudgingly delivers.

That is bad administrative law and bad policy.