Analyze NHTSA Vehicle Safety Complaints
Collect de-identified NHTSA complaint narratives and components to research recurring vehicle safety signals.

Consumer complaint data can reveal recurring language before a pattern becomes obvious in aggregate reports. It can also mislead when analysts treat every narrative as a verified defect. The NHTSA Vehicle Safety Scraper collects official complaint records by vehicle configuration while deliberately excluding VINs.
Collect complaint signals
Turn on complaints and disable recalls when the analysis only needs reports:
{"vehicles":[{"make":"Tesla","model":"Model 3","modelYear":2022}],"includeRecalls":false,"includeComplaints":true,"maxResultsPerVehicle":500}Rows include complaint number, incident and filing dates, component, summary, crash and fire indicators, injury and fatality counts, mileage when available, and the queried make, model, and year. Retain the complaint number for deduplication. Narrative text may contain spelling differences, incomplete descriptions, or repeated language, so preserve the raw summary and add derived labels in separate fields.
A useful first pass groups reports by component and month, then measures the share mentioning a crash, fire, injury, or fatality. Text clustering can surface repeated phrases, but sample size must remain visible. Ten similar reports do not establish a failure rate because the public dataset does not provide the number of vehicles exposed, reporting propensity, or an investigation outcome.
Compare complaint patterns with official recalls and manufacturer communications, and inspect the underlying records before publishing conclusions. Sudden changes may reflect publicity, model-name matching, or reporting behavior rather than a new mechanical event. Avoid ranking models on raw complaint totals without accounting for sales volume, age, and time on the road.
The Actor's VIN exclusion supports data minimization, but teams should still govern narrative access because free text can occasionally contain details supplied by a complainant. Use the dataset for trend discovery and research triage, not personal profiling or definitive safety claims. Cite NHTSA as the source, state the collection date, and make the distinction between reported experience, investigation, and confirmed recall unmistakable.
Frequently asked questions
Are complaints proof that a vehicle has a defect?
No. Complaints are reports and should be treated as signals that require validation.
Does the output contain VINs?
No. The Actor deliberately excludes VINs from complaint output.
Related
100 free credits, no credit card.
About 30 real searches. Add the MCP to Claude or Cursor in two minutes.