The record isn't open yet. We're collecting the first accounts now — report before it opens and you get six months of free access when it does.
wayward fowlbeta
Building the public record of AI harm

A computer decided
something about you.
Nobody will explain it.

Denied cover. Refused credit. Screened out of a job before a person read your name. Flagged, scored, cut off — by software you never saw, on grounds nobody will put in writing.

We are collecting these, one account at a time, and holding them to a standard that stands up. Tell us what happened to you.

Step 1 of 5

What happened?

Do you know what system made the decision?
Most people don't, and it doesn't hold the report up. Working it out is our job.
Who made the decision? (if you know)
The insurer, the bank, the utility, the employer, the agency.
Tell us what happened
In your own words. What was decided, when, and what it meant for you. No legal language needed.

Our promise to you

A record of one report proves nothing. A record of a thousand, each held to the same standard, is a pattern an organisation has to answer for. That is the whole point, and it only works if people come forward before there is anything to see.

We do something with it

When enough people describe the same thing happening at the same company, we take it to them — as an anonymised, aggregated pattern, never as your name and never as your account on its own. Then we push to get it put right, and we publish what they say back.

You get the tool

Everyone who reports before the record opens gets six months of free access to all of it — the reports, the patterns, and the data behind them. You helped build it. You should be able to use it.