Trust Center
RecallIQ is an independent consumer information service. We are not a government agency and we do not issue recalls. Federal agencies do. Our job is to make what they publish findable, understandable and actionable.
Our sources
Every published recall record traces to an authoritative federal source, and every recall page links to the original announcement. We separate sources into two lanes.
Public alert lane
Used for consumer-facing recall pages and alerts.
CPSC — CPSC Recalls API
Consumer Product Safety Commission recall announcements for consumer products, toys, appliances and children's products.
Access: Official REST API · Checked every 60 minutes · Attribution: U.S. Consumer Product Safety Commission
FDA — FDA Enforcement Reports
FDA enforcement report dataset covering food, drug and device recall classifications.
Access: Official open data API · Checked every 60 minutes · Attribution: U.S. Food and Drug Administration
FSIS — USDA FSIS Recalls
Meat, poultry and egg product recalls and public health alerts.
Access: Official JSON feed · Checked every 60 minutes · Attribution: USDA Food Safety and Inspection Service
NHTSA — NHTSA Vehicle Recalls
Vehicle safety recalls, including VIN and year/make/model lookups.
Access: Official REST API · Checked every 180 minutes · Attribution: National Highway Traffic Safety Administration
Research lane
Used only for research, context and enrichment. A record from this lane can never trigger a consumer alert or become a recall page.
FDA — FAERS Adverse Events
Adverse event reports. Research and enrichment only — never used to issue a consumer alert.
How recall information is collected
Each agency has its own connector that runs on a schedule. Records are stored exactly as received before anything is done to them, then validated, normalized into one schema, checked for duplicates and updates, enriched, fact-checked and only then published.
source → raw store → validate → normalize → deduplicate → verify → enrich → fact-check → review → publish
How AI is used
AI never decides that something has been recalled and never invents recall facts. It works only from the verified source record.
AI may
- Write a plain-language summary
- Categorize and tag a record
- Extract structured identifiers
- Draft page metadata and alert copy
- Flag likely duplicates for a human
AI may not
- Add facts the source does not contain
- Change official consumer instructions
- Assign or inflate severity
- Invent lot, model or geographic details
- Turn a complaint or rumor into a recall
What a human reviews
Automated fact-checking compares every generated sentence against the source record. Any material discrepancy stops publication. Records involving death, serious injury, infants, children's products, medication, medical devices, severe allergens, conflicting sources or low confidence always reach a human reviewer before they publish.
Corrections policy
We never silently change safety information. Every recall page shows a last-updated date, we keep a full version history of each record, and significant corrections are labelled on the page itself. If you spot an error, tell us and we will check it against the agency record.
Advertising and affiliate policy
Advertising is off by default while we build trust. When advertising is enabled, it will never appear inside hazard information, consumer instructions, remedy details, company contact information or the official source link, and it will never be styled to look like a recall notice or a government message. Advertisers and affiliates have no influence on which recalls we publish or how they rank.
What “no results” means
If a search returns nothing, it means no verified recall in our database matched what you typed. It does not mean the product is safe. Recalls are issued over time, and identifiers are sometimes recorded differently than they appear on packaging.