Recall Check for Resellers (CPSC, NHTSA, FDA)

Checks a list of items (title, brand, model, UPC) against US recalls from CPSC, NHTSA and FDA and returns one row per item with a match status, the recall details and a do_not_sell flag.

It's illegal to sell a consumer product that's under a publicized recall. The law (15 U.S.C. 2068) covers anyone who would "sell, offer for sale … or distribute in commerce" a recalled product, and that includes resellers. This Actor does the tedious part: it matches your inventory against every recall on record and tells you which items to pull.

Who it's for

• Thrift, consignment and resale shops
• Auction houses and estate-sale companies
• Liquidation and returns-pallet buyers, bin stores
• eBay, Amazon, Facebook Marketplace and Mercari sellers
• AI agents that list or price second-hand goods

What you get

One row per input item:

Field	Meaning
match_status	match, possible_match or no_match
do_not_sell	true only for confirmed matches (confidence ≥ 0.9)
confidence	0–1 score for the best recall found
agency, recall_id, recall_title, recall_date	Which recall, from which agency
hazard, remedy, recall_url	What's wrong, what the maker offers, and the official notice
sold_at	Where and when the recalled product was sold (CPSC), or where it was distributed (FDA)
caveat	What to check before deciding (model numbers, lot codes, VINs, dates)
matched_fields, match_reason	The evidence used, e.g. upc; brand and a plain-English reason
candidate_count, other_recall_ids	Other recalls that also scored above your threshold
agencies_checked, source_url, retrieved_at	Where the data came from and when it was pulled
item_*	Your input echoed back, including your own item_id (SKU or lot number)

How matching works

1. UPC listed in the recall → match
2. Model number listed in the recall + same brand → match
3. Brand + product name, when the recall covers all models (for example, all Rock 'n Play sleepers) → match
4. Model number without a brand, a close model number, or brand + a similar product name → possible_match
5. A similar product name only → weak possible_match

If a recall lists specific models or UPCs and yours isn't among them, the score drops. Vehicle recalls are never more than possible_match, because they cover specific VINs.

Input

Give items inline, as a CSV/JSON link (Google Sheets share links work), or as an Apify dataset ID. Spreadsheet exports work as they are: an auction catalog with Lot #, Lead and Description columns, an eBay or Shopify export, or your own inventory sheet. Long descriptions are searched for brands, model numbers and UPCs.

{
  "items": [
    { "id": "LOT-101", "title": "Hampton Bay Halwin 52 in. ceiling fan", "brand": "Hampton Bay", "model": "AK396H-MBK" },
    { "id": "LOT-102", "title": "Fisher-Price Rock 'n Play Sleeper", "brand": "Fisher-Price" },
    { "id": "LOT-103", "title": "Lowes Foods sour cream and onion potato chips", "upc": "741643055766" },
    { "id": "LOT-104", "title": "2021 Volvo XC60", "year": 2021 }
  ],
  "agencies": ["CPSC", "NHTSA", "FDA"],
  "min_confidence": 0.6,
  "since_year": 1970
}

Input	Type	Default	Notes
items	array	4 demo items	Each needs at least one of title, brand, model, upc. Optional: description, category, year (vehicles), id. Common spreadsheet headers are understood too, and a plain string counts as a title. Pass UPCs as strings.
items_url	string	empty	CSV or JSON. Recognized columns include title/name/lead, description, brand/manufacturer/make, model/model number, upc/barcode/gtin, year, id/sku/lot.
dataset_id	string	empty	Read items from another Actor's dataset.
agencies	array	all three	CPSC (consumer products), NHTSA (vehicles, car seats, tires, equipment), FDA (food, drugs).
min_confidence	number	0.6	Ignore weaker candidates.
since_year	integer	1970	The full history is checked by default, because old recalled products still show up at resale.

Output example

{
  "item_index": 0,
  "item_id": "LOT-101",
  "item_title": "Hampton Bay Halwin 52 in. ceiling fan",
  "item_brand": "Hampton Bay",
  "item_model": "AK396H-MBK",
  "item_upc": null,
  "match_status": "match",
  "do_not_sell": true,
  "confidence": 0.94,
  "agency": "CPSC",
  "recall_id": "26702",
  "recall_title": "Hampton Bay Halwin 52-Inch Ceiling Fans Recalled Due to Impact and Injury Hazards; Manufactured by Youngo Limited",
  "recall_date": "2026-08-13",
  "hazard": "The fan blades can separate from the fan motor flywheel, posing impact and injury hazards to consumers.",
  "remedy": "Consumers should stop using the recalled ceiling fans immediately and contact Youngo for either a full refund…",
  "recall_url": "https://www.cpsc.gov/Recalls/2026/Hampton-Bay-Halwin-52-Inch-Ceiling-Fans-Recalled-Due-to-Impact-and-Injury-Hazards-Manufactured-by-Youngo-Limited",
  "sold_at": "The Home Depot stores nationwide and online from January 2023 through October 2025 for about $180.",
  "caveat": "Check the model numbers, dates and labels in the recall notice before deciding.",
  "matched_fields": "model; brand",
  "match_reason": "Model AK396H-MBK and brand Hampton Bay match this recall.",
  "candidate_count": 2,
  "other_recall_ids": "CPSC:21059",
  "agencies_checked": "CPSC; NHTSA; FDA",
  "error": null,
  "source_url": "https://www.saferproducts.gov/RestWebServices/Recall?format=json&RecallNumber=26702",
  "retrieved_at": "2026-09-30T06:00:00Z"
}

Pricing

Pay per event: $0.003 per item checked (match or not). There's no extra charge for matches. Items with nothing to check (no title, brand, model or UPC) are returned with an error and are not charged. The Actor stops cleanly at your maximum cost per run.

Data sources

All official US government data, used under their public terms:

• CPSC recalls via the SaferProducts.gov Recall Retrieval web service (1973 to today).
• NHTSA recall flat files covering vehicles, equipment, tires and child car seats.
• FDA enforcement reports (food and drugs) via openFDA bulk downloads. Data provided by the U.S. Food and Drug Administration (open.fda.gov).

The data is rebuilt every day, and a new build goes live only after it passes a self-test against known recalls. CPSC recalls published since the last rebuild are added at the start of every run. Each result row shows when its data was retrieved (retrieved_at).

Limits

• "No match" is not a safety guarantee. Recall records can be incomplete, and older recalls rarely list UPCs. Always open the recall link and compare model numbers, dates and labels.
• Vehicle recalls apply to specific VINs. Check the VIN at nhtsa.gov/recalls.
• FDA recalls usually cover specific lots or dates. Compare the lot code on the package with the caveat text.
• FDA medical-device recalls aren't included yet.
• US recalls only.

Disclaimer

• Provided "as is." This Actor and its results come with no warranty of any kind, express or implied, including accuracy, completeness, fitness for a particular purpose or non-infringement. You're responsible for your own decisions about what to sell.
• Not affiliated with the government. Robot Supply Store is not affiliated with, endorsed by or sponsored by the U.S. Consumer Product Safety Commission (CPSC), the National Highway Traffic Safety Administration (NHTSA) or the U.S. Food and Drug Administration (FDA). Their names describe where the public data comes from.
• Not legal or medical advice. Results support your own compliance checks. For a recalled product, follow the recall notice and contact the recalling firm or the agency.

FAQ

Is a possible_match safe to sell? Not until you've checked. Open recall_url and compare the model number, date codes and description.

Why did a well-known recall come back as possible_match? Usually the input had no model number or UPC. Add them and run again.

Can an AI agent call this? Yes. It's pay-per-event, so it's available through Apify's MCP server. Every field is flat JSON with stable names.

Is this legal advice? No. It's data to support your own compliance checks.