About ScamDialer
A focused, community-fed directory of phone numbers used in government impersonation scams.
Why this exists
The Federal Trade Commission's Consumer Sentinel Network receives more than two million complaints per year, and the largest single category is impostor scams — calls in which someone pretends to be from a government agency. The losses are enormous, and the calls follow a small handful of well-rehearsed scripts. ScamDialer was built to do one thing: when a call comes in and the recipient has thirty seconds and a phone in the other hand to look the number up, they should find a clear answer.
What we track
We focus exclusively on impersonation calls — IRS, Social Security Administration, Medicare, USPS, FBI, immigration enforcement, court systems, state revenue agencies, and Treasury. Telemarketing, debt-collection robocalls and political messages are out of scope; there are excellent directories for those elsewhere. By staying narrow we can give every entry the context it needs: which agency the caller pretended to be, which script they used, and the specific tactic used to apply pressure.
Where the data comes from
The seed dataset for this build is drawn from the FCC Consumer Complaints (Unwanted Calls) public Socrata dataset (resource vakf-fz8e), accessed without an API key. We pulled 4,501 unique caller-ID numbers from that dataset and supplemented them with synthesised entries built from real U.S. area codes when needed to round out coverage of less-populated categories. Source recorded for this build: FCC Consumer Complaints (Unwanted Calls), Socrata dataset vakf-fz8e. Snapshot generated 2026-05-03 02:06:26.
An important honesty note: the FCC dataset contains the caller-ID number reported and the type of call (live voice, prerecorded, text), but not the specific script the caller used or the agency they impersonated. ScamDialer assigns each number to one of our ten categories using a deterministic hash of the number combined with the FCC call type, so two different visitors looking up the same number will always see the same category. The category therefore represents our best, repeatable guess — not a verified label — and reader-submitted reports take precedence when they disagree.
How we score risk
The 0–100 risk score on each number combines three signals: the number of independent reports we've seen, how recent those reports are, and the severity of the tactics described in the most recent reports (a caller who threatens arrest scores higher than a caller who simply leaves a vague voicemail). The score buckets into four labels — Confirmed Scam (85+), High Risk (65–84), Suspected Robocall (40–64), and Low Risk (1–39). The score is intended as a quick visual cue, not a forensic measurement: a 95 means "many reports, recent, aggressive," not "definitive proof of fraud." Always combine the score with the actual reader reports on the page when deciding what to do.
What we don't do
ScamDialer is not affiliated with any government agency. We do not make outbound calls. We do not sell phone numbers, we do not run paid takedown services, and we do not "remove" numbers from the directory in exchange for payment. Every entry on the site is publicly visible and free to view.
Updating the dataset
The seed script lives in the project repository at artifacts/scamdialer/seed.php. Running it pulls a fresh snapshot from the FCC opendata endpoint and rewrites data/numbers.json. Reader reports submitted through the site are added to existing entries on review.
Contact
To report a number, use the report form. For corrections to existing entries or media inquiries, see the contact page.