About.

What this is.

Honest Privacy Policies is an editorial project that reads the privacy policy of every company you tap "I agree" on and translates it into the version your most cynical friend would write.

For every company you'll find a grade, a Privacy Label (front and back), eight plain-English answers to the things people actually want to know, marked-up excerpts of the worst clauses, the dark patterns we caught, the rights you have where you live, and — most importantly — receipts: every claim above links to the line in the policy we got it from, both at the original URL and in a snapshot of what the policy said the day we read it.

Why it exists.

Nobody reads privacy policies, and the people who write them know that. The policies are written to be defensible in court, not understandable to humans. The result is a quiet, ambient deal where companies extract value from your behavior in exchange for products you didn't realize you were paying for.

We can't fix that with one website. We can at least make it cheap to find out what you agreed to.

How it works.

1. We fetch the policy and take a hash of it so we can prove what it said.
2. A large language model (currently Claude) reads it under a versioned prompt and emits a structured JSON: grade, label, big questions, marked excerpts, dark patterns, rights, sources.
3. A human reviews the output before it's published.
4. Every claim has a section reference that links back to the original line in the policy and to our local snapshot of that day's version.
5. If the policy changes, we re-fetch, re-hash, re-analyze, and keep the diff.