Coming to iOS · U.S. launch list open

Dashcam-documented driving incidents.

Record a 60-second clip. BadDrivr's on-device ML handles plate, vehicle, and incident detection — then files a structured, shareable record with moderation and driver right-of-reply built in from day one.

Six on-device classifiersEight incident typesDriver reply by design
Submitted · 2 min agoreport · 9k3xR
FLABC-1234Tailgating

Biscayne Blvd, Miami, FL

0:47 clip · auto-detected: white Tesla Model 3

Plate · FL98%Make · Tesla94%Color · white91%
14 Hawks2 comments2d ago
Vehicle owner can claim this plate to post a public reply.

What's in it

Documentation, not a rant thread.

BadDrivr is built to turn a brief, chaotic moment into something structured enough to be useful — to you, to a vehicle owner who wants to reply, and to a community that can flag, surface, or challenge any single report.

On-device ML

Your iPhone does the detection.

Six on-device classifiers handle plate text, state, vehicle make, type, color, and badge readout — six pipelines running locally before a single byte uploads.

Structured, not freeform

Eight incident categories.

Tailgating, cut off, no signal, bad parking, reckless, distracted, left-lane camper, parking-lot menace. A structured category, not a caption.

Driver right-of-reply

Two sides on every record.

Vehicle owners can claim their plate and post a public 300-character reply to any report. One side of the record, then the other — visible to every reader.

Moderated from day one

Automated + human review.

Content checks run on device and server-side. Flagged content and appeals go to human moderators. Zero-tolerance categories escalate immediately. 48-hour SLA for NCII takedowns under the TAKE IT DOWN Act.

Searchable public record

Every report has a URL.

Reports live at shareable links. Search plates by prefix, browse cities by engagement, explore category leaderboards — on iOS and on the web.

Privacy by default

No names. No DMV. No addresses.

We don't identify drivers by name, we don't query DMV records, and we don't display exact street addresses — only city-level location. U.S. focus at launch.

How it works

From a 60-second clip to a structured record.

  1. 01

    Record

    A dashcam or iPhone captures up to 60 seconds of the incident. You keep the original file — we don't access anything you don't upload.

  2. 02

    Auto-tag

    Six on-device ML classifiers prefill the plate, state, vehicle make, color, and type. You review, pick an incident category, and submit.

  3. 03

    The record lands

    The report gets a shareable URL and enters the moderated public record. The vehicle owner can claim the plate to post a public reply; anyone can flag or request review.

Built with moderation in mind

We don't verify every report. We give every report a way to be challenged.

BadDrivr is a community reporting platform — not a court, not a law enforcement tool, not a driver-identification service. Reports are user-submitted observations that can be restricted, removed, or corrected through moderation and takedown channels.

Both users and non-users can challenge content. The public site lists every channel; the iOS app has in-product flag, appeal, and claim flows. None of this is optional — it ships in the first release.

  • U.S. focus at launch.
  • No driver identification by name.
  • No DMV or registration lookups.
  • No exact home addresses on public pages — city-level only.
  • Non-users can request review of a report without creating an account.
  • Zero-tolerance categories (CSAM, NCII, credible threats) removed on detection.
  • Every report has a takedown path at /takedown.
  • We do not sell personal information.

Launch list

Be there when it goes live.

BadDrivr is in final pre-launch review for iOS in the United States. Drop a note and we'll send a single email when the app is live on the App Store. No spam, no autoresponders — it's a small team and every message gets a reply from a human.

Email the launch deskben@baddrivr.com · reviewed by a human