Fake number detector
Paste or upload a batch of numbers and see whether they look genuine or made-up, using Benford’s Law. Get a clear verdict, a suspicion score, a chi-square p-value and a digit chart — with first-digit and first-two-digit tests, all in your browser.
What is Benford’s Law?
In real-world numbers that span many sizes — tax amounts, incomes, expenses, populations — the first digit is not spread evenly. A “1” leads about 30% of the time, while a “9” leads under 5%. The exact pattern is P(d) = log10(1 + 1/d):
| Leading digit | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 |
|---|---|---|---|---|---|---|---|---|---|
| Expected | 30.1% | 17.6% | 12.5% | 9.7% | 7.9% | 6.7% | 5.8% | 5.1% | 4.6% |
When people invent numbers, they spread the digits too evenly, shy away from leading 1s, and lean on 5s and round figures. That break from the curve is the tell. This tool measures how far your numbers stray from the expected pattern and turns it into a verdict.
Important: Benford’s Law works on a large batch of numbers, never a single value, and a deviation is a red flag to investigate, not proof. It needs natural quantities that span several sizes — it does not apply to assigned numbers like IDs, phone numbers or years, or values stuck in a narrow range.
Files are processed in your browser and never uploaded.
How to use Fake number detector
Add your text
Type or paste your text into the input box. Everything is handled by your own browser, so nothing is sent to a server.
Run it
Press Check numbers, Don → to produce the result.
Start over any time
Nothing is saved on a server, so reloading the page clears everything.
What this tool does
First-digit and first-two-digit Benford's Law tests
Benford's Law says real-world figures start with 1 far more often than with 9. Both the first-digit and the first-two-digit tests are run against your data.
Sample-aware verdict using a chi-square p-value
The verdict takes the sample size into account through a chi-square p-value, so forty numbers are not judged as if they were four thousand.
Suspicion score from 0 to 100
The suspicion score condenses the tests into a single number from 0 to 100, for when you want one answer rather than a table.
Mean Absolute Deviation (MAD) and chi-square
Mean Absolute Deviation and the chi-square statistic are both shown, so the result can be checked rather than taken on trust.
Round-number and duplicate-value checks
Round numbers and repeated values are counted separately, because invented figures cluster on them in a way real ones do not.
Paste or upload CSV/TXT, runs on your device, nothing uploaded
Paste the numbers or load a CSV or TXT file. Everything is calculated on your device and nothing is uploaded.
Frequently asked questions
What does this tool do?
It checks a batch of numbers against Benford’s Law, which describes how often each digit naturally appears first in real-world data. If your numbers stray far from that pattern, it flags them as possibly fabricated and shows how far off they are.
Can it prove a number is fake?
No. Benford’s Law is a red flag, not proof. A suspicious result means the batch is worth investigating against the real records, not that any single number is definitely false.
How many numbers do I need?
The more the better. Aim for at least 50, ideally hundreds, and they should span several sizes (small and large amounts). Benford works on a population, not one value.
What kinds of numbers does it not work on?
Assigned or bounded numbers do not follow Benford's Law, such as ID numbers, phone numbers, postal codes, years, or values stuck in a narrow range like ages or shoe sizes. Use natural quantities like amounts, totals or counts.
What is Simulate Natural Variation?
It rewrites the leading digits of your batch so the numbers follow Benford’s Law, then lets you re-check them with one click. It’s a quick way to see why a Benford result is a red flag and not proof, and it’s handy for generating Benford-style test data.
Does it work on small batches?
The verdict is based on a chi-square p-value, which adjusts for how many numbers you have, so genuine data isn’t unfairly flagged when you only have a few hundred values. More numbers still give a firmer read.
Is my data uploaded?
No. Everything is parsed and analysed in your browser on your own device. Your numbers never leave your computer or phone.
Last updated: