Recall Forensics

Recall Forensics: The Darwin's DOJ Injunction Lawsuit — the validation failure behind it

Team ValidLeaf · October 2026 · 7 min read

Recall Forensics is our running teardown of real pet food enforcement actions: we pull the public regulator records and trace each case down to the exact quality-system weakness underneath it. Earlier this month we covered the Darwin's recall that never happened — six pathogen-positive lots, an FDA recall recommendation, and no recall. That story now has a second chapter, playing out in federal court. The question is no longer whether the company will pull product. It is whether it will be allowed to keep making it — and the answer will turn on records: whether its preventive controls can be proven to work, not merely described.

What happened

On August 7, 2026, the U.S. Department of Justice, acting on behalf of the FDA, filed a civil complaint in the U.S. District Court for the Western District of Washington (Case No. 2:26-cv-02805) seeking a permanent injunction against Arrow Reliance Inc., doing business as Darwin's Natural Pet Products, and its founder and president, Gary T. Tashjian. The government asks the court to bar the defendants from manufacturing, processing, packaging, holding, or distributing animal food unless their operations comply with federal food safety requirements to the FDA's satisfaction — and to give FDA ongoing inspection authority to verify that compliance (Vet Candy, Oct 24, 2026; Petfood Industry).

The complaint's most quoted figure is a sampling record. Of the 18 most recent samples of Darwin's raw pet food collected by FDA since 2023, the government alleges 12 contained pathogens: nine with Salmonella, one with Listeria monocytogenes, and two with both. Third-party testing of unopened product in 2025 reportedly found Salmonella in two lots and Shiga toxin-producing E. coli (STEC) in another. And the scale is not small: roughly 85,000 pounds of frozen raw pet food per week from the Tukwila, Washington facility, shipped directly to consumers and through third-party warehouses (Petfood Industry; Vet Candy).

The complaint also describes the human case that raises the stakes beyond pet illness. In 2024, a four-year-old child developed an E. coli O157:H7 infection, hemolytic uremic syndrome, and chronic kidney disease. The family's dog ate Darwin's raw food exclusively and began vomiting a day before the child fell ill; whole genome sequencing found the E. coli from an unopened package was an exact genetic match to the isolate from the child, according to the complaint. From 2016 through January 2026, FDA also received 49 consumer complaints reporting illnesses involving 34 dogs (five reported deaths) and 35 cats (four reported deaths) fed the company's products (Petfood Industry; Vet Candy).

Then there is the plant itself. Describing a 2024 inspection, the complaint alleges FDA found Salmonella inside a hopper used to package finished product; condensation dripping onto sanitized equipment; aerosolized wash water blowing onto exposed finished food; pitted and rusted food-contact equipment; deeply cut cutting boards; damaged flooring pooling water, blood, and raw ingredients; and employees touching unsanitized surfaces and then handling food. It further alleges the facility was never registered despite a 2023 FDA warning that registration was required, and that website claims linking the diets to diabetes, arthritis, seizures, urinary tract infections, and IBD made the products unapproved animal drugs (Petfood Industry; Vet Candy).

The company is fighting back. In an October 13, 2026 statement, Tashjian said the company disputes the FDA's interpretation of federal requirements, arguing it is inconsistent with the science of raw pet food and stricter on Salmonella than the approach to some raw meat sold for human consumption. The company notes it has not been ordered to stop operations, its products remain available, and it has moved to dismiss two of the complaint's six counts. As of October 24 docket reporting, the case remains in litigation, with a joint status report scheduled for October 30, 2026 (Vet Candy).

An honesty note: these are allegations in a complaint, not findings by a court. No ruling establishing liability, and no order stopping production, has been issued; the company disputes the conclusions and says its practices meet applicable requirements. What is not in dispute is the escalation itself — a decade of sampling, advisories, warning letters, and declined recall requests has become an injunction case, the most serious tool in the federal food-safety kit.

The validation gap: controls that could not prove they worked

Our earlier Darwin's post was about a recall plan nobody executed. This one is about something deeper, and it is the lesson every pet food manufacturer should take from the complaint. According to the government, the company had identified Salmonella, E. coli, and Listeria as potential hazards in its own analysis — and used measures including peracetic acid treatment and cold storage — yet pathogens allegedly kept appearing in finished product. Naming a hazard is the first page of a food safety plan under the animal food preventive controls rule (21 CFR Part 507). It is not control. Control is a measure, applied at a defined point, with defined parameters, monitored, and — critically — validated and verified with records showing it actually reduces the hazard it claims to reduce. When a control keeps failing in finished-product testing and the plan does not change, what you have is not a preventive control. It is a description of one.

The facility findings, if proven, point at the second half of the same gap: equipment and sanitation are qualification questions. A food-contact surface that is pitted, rusted, or carved with deep grooves cannot be reliably cleaned — that is not a housekeeping failure, it is a cleanability failure designed into the line. Wash water aerosolized onto exposed product and condensation falling onto sanitized equipment are cross-contamination routes that a sanitation program must be designed and verified against, not just scheduled around. And Salmonella detected inside the packaging hopper for finished product is exactly what an environmental monitoring program exists to surface early — in the plant's own data, on the plant's own schedule, long before a regulator's sample does it in a court filing. Compare the Lewis Sausage warning letter posted earlier this month: a different company, a different product, the same underlying pattern — a hazard analysis that named the pathogens and validation studies that never proved the reduction.

The claims count carries a change-control lesson. Language suggesting a food can improve or resolve diabetes, seizures, or urinary tract disease is not marketing decoration; regulators treat disease claims as evidence the product is intended as a drug, with an approval pathway it never went through. In a disciplined quality system, web copy passes the same review gate as a formulation change. In an undisciplined one, marketing publishes first and regulatory finds out from a complaint — literally, in this case, a federal one.

Finally, notice what an injunction would demand: not a fine, but production stopped until compliance is demonstrated to the FDA's satisfaction, with inspectors authorized to keep checking. That is the end state of every unresolved finding — the burden of proof moves. Today a manufacturer asserts its controls work; under an injunction it must prove they work, continuously, to the agency that sued it. The cheapest time to build that proof file is years before anyone files a complaint.

How a GMP Readiness Assessment catches it

This is exactly the failure pattern a GMP Readiness Assessment is built to find while findings are still cheap to fix, in pet food operations governed by 21 CFR Part 507. Over four weeks, it stress-tests:

• Preventive-control evidence, not just the plan — for every hazard your analysis names, is there a validation study showing the control reduces it, monitoring records showing it ran, and verification showing someone checked? A control with no evidence file is scored as a gap — a regulator will score it that way too.
• Environmental monitoring that trends — are food-contact and near-food-contact surfaces sampled on a risk-based schedule, are positives investigated to root cause, and would your own data find a pathogen in a packaging hopper before an inspector's swab does?
• Equipment condition and cleanability — pitted, rusted, or deeply scored food-contact surfaces, damaged floors, condensation and wash-water routes onto exposed product: the plant audited as a qualification problem, not a janitorial one.
• Claims and label change control — do website, label, and marketing claims pass a regulatory review gate, so a disease claim can never quietly convert a food into an unapproved drug?
• Finding-closure discipline — every observation, complaint trend, and positive result tracked to a CAPA with an effectiveness check, so a ten-year trail of open findings can never accumulate.

It closes with a gap assessment, a remediation roadmap, and starter templates — so the proof file exists before anyone needs it as evidence.