Recall Forensics

Recall Forensics: The Brad's Kale Chips Thallium Recall — the validation failure behind it

Team ValidLeaf · November 2026 · 6 min read

Recall Forensics is our series tracing real product recalls to the specific validation gap behind them. Every fact comes from public regulator records or reporting of those records — no speculation, no invented causes. This case starts in the ground, not in the plant. On November 2, 2026, a Florida snack maker recalled about 25,000 bags of kale chips because the kale itself may have carried thallium, a metal it drew up from the soil — and dehydration, the one process step the product does have, made the concentration worse, not better.

What happened

Snack Right Super Foods, Inc., of Coral Gables, Florida — doing business as Brad's Raw Chips — announced the voluntary recall on November 2, 2026, and the U.S. Food and Drug Administration published the notice on November 3. It covers four Brad's Plant Based products: Crunchy Kale in Original, Cheeze It Up, and Radical Ranch (2 oz bags) and Veggie Chips Kale (3 oz bags). About 25,000 bags, roughly 2,097 cases, are estimated to be in commerce. They were distributed nationwide through KeHE Distributors and United Natural Foods, Inc. (UNFI), and sold directly to five individual consumers, with distribution between October and December 2025. Best-by dates on the recalled lots run from December 21, 2026 to February 17, 2027. Consumers were told to stop eating the products and discard them; refunds are not being offered, and the products should not be returned to stores (FDA recall notice, November 3, 2026; Reader's Digest, November 4, 2026).

The hazard is chemical, not microbial. The FDA notice describes thallium as a naturally occurring metal found in soil and in sulfur-based minerals, which plants and vegetables can take up — particularly members of the Brassicaceous family, the cabbage family to which kale belongs. At relatively high levels, acute or chronic exposure can cause gastroenteritis, damage to peripheral nerves, and alopecia, or hair loss. Two sentences in the notice carry the whole quality lesson. The company states that its manufacturing process did not introduce the contaminant — and that dehydration removes approximately 85 to 90 percent of the product's water, which concentrates elements already present in the fresh kale (FDA).

The signal also did not start in November. In May 2026, the California Department of Public Health advised consumers not to eat certain Brad's Plant Based products after it investigated an inquiry from one individual with elevated thallium levels. California officials collected samples at retail stores in the Sacramento area, and laboratory testing found elevated thallium levels in multiple varieties. The November recall was initiated after FDA advised the firm of thallium in tested products (FDA).

An honesty note, because we hold ourselves to it: the FDA notice does not publish the thallium levels measured in the products, does not name the farm, growing region, or kale supplier involved, and does not report illnesses beyond the California inquiry into one individual with elevated levels that began the investigation. Nothing below treats a specific level, a specific field, or a specific supplier failure as established fact. The documented facts are the uptake pathway the notice describes, the concentration effect of dehydration, the May state advisory, and the November recall.

The validation gap: when there is no kill step, the field is part of the process

Most food-safety systems lean, quietly, on a downstream step that can rescue an upstream problem: a cook that kills pathogens, a wash that reduces them, a filter that removes something. A raw, dehydrated kale chip has no such step for a metal. Nothing in the process removes thallium, and the defining process step — drying — concentrates whatever the leaf carried in. That single fact rearranges where validation has to live. If the hazard arrives with the crop and the process amplifies it, the preventive control cannot be in the plant. It has to be in the hazard analysis, the sourcing decision, and the incoming-material specification.

• Name soil-borne metals in the hazard analysis for accumulator crops. A hazard analysis for leafy brassicas that covers pathogens and pesticides but never asks what the crop pulls from its soil is incomplete in a predictable way. Kale's family is known, in the FDA notice's own words, to take up elements from soil. The analysis should treat heavy-metal uptake as a reasonably foreseeable chemical hazard for accumulator crops — with the growing region and supplier as part of the assessment, not as purchasing trivia.
• Do the concentration math during process validation. Process validation usually asks what a step does to microbes, moisture, and shelf life. For a dehydrated raw product it must also ask what the step does to hazard concentration. A process that removes 85 to 90 percent of the water leaves the solids — and everything dissolved or held in them — in a much smaller mass. A level that looks unremarkable in fresh kale is not the same level in the finished chip. Specifications set on the fresh ingredient without that concentration factor are specifications for a product you do not sell.
• Put a metal specification on the incoming crop, and test against it. For an ingredient with no downstream removal step, the receiving dock is the control point. That means a written specification for relevant metals in incoming kale, a defined sampling and testing cadence scaled to supplier and region, and certificates of analysis that are verified periodically rather than filed. A supplier programme that qualifies a grower on food-safety paperwork alone, with no analytical verification of the hazard the crop is known to accumulate, is qualification in name only.
• Treat an outside agency's advisory as a quality signal with a clock on it. A state health department told consumers in May 2026 not to eat certain products from this line, after finding elevated levels in multiple varieties at retail. The federal recall followed in November. Whatever internal work happened in between is not public, and we do not speculate about it. The general lesson stands on the dates alone: an external finding about your product is not a press problem to monitor — it is nonconforming-signal data that belongs in the quality system, with a documented investigation, retain-sample testing, and a hold-and-scope decision that can be shown later.
• Keep retain samples that can answer a chemical question months later. The recalled bags were distributed between October and December 2025, with best-by dates into early 2027. A signal that surfaces the following spring can only be bounded if retained product still lets the firm test what actually shipped, lot by lot. Without that, the honest scope is broad.

We have traced the "no kill step" family before, in a different key: the Salata dressing recall, where a raw ingredient carried a pathogen into a product with no later step to remove it. This case extends the logic from biology to chemistry — and adds the twist that the process does not merely fail to remove the hazard, it concentrates it. For food and beverage manufacturers building around raw or dried produce, the question to answer in writing is: which hazards does our crop bring with it, what does our process do to their concentration, and where — exactly — is the tested control?

How a GMP Readiness Assessment catches it

This is exactly the failure pattern a GMP Readiness Assessment is built to find. Over four weeks, it reviews:

• Hazard analysis completeness for chemical hazards — does the analysis for each crop consider soil-borne metals and other uptake hazards, or does it stop at pathogens and pesticides?
• Concentration effects in process validation — has anyone documented what dehydration, reduction, or evaporation steps do to the concentration of hazards already in the ingredient?
• Incoming-material specifications and verification testing — do high-risk crops carry analytical specifications, and are supplier results verified on a defined cadence instead of filed?
• Grower and supplier qualification depth — does qualification reach growing region and practice, or only the packer's paperwork?
• External-signal handling — is there a written route by which a state advisory, a regulator's test result, or a single elevated-level inquiry becomes a tracked investigation with a documented decision?

It closes with a gap assessment, a remediation roadmap, and starter IQ/OQ/PQ templates — so the fix starts before the lab report does.