Unmatched AI Disclosure
Last updated: November 17, 2026.
Unmatched uses an AI service to read your dating profile and write your critique. This page says what goes to that service, what does not, and what the output is.
What goes to Anthropic's Claude
Each time you submit a profile for an issue, our server sends the following to Anthropic, the company that makes Claude, through its commercial API. Nothing is sent when you create an account or open the app. Nothing is sent when you browse old issues. The AI is also used later, if you take the optional interview or ask for rewrites, and those steps send text as described below.
- The screenshots you picked, which include the text that appears in them.
- Your match preference (women, men or everyone) and your age range, if you set them, as one line of context for the critique.
- Your mood setting (honest or savage) and the dating app the profile is from.
- For the interview and rewrites, text only: the AI-written descriptions of your weak prompts and bio, the editor notes on them, and your interview answers, if you give any.
Your match preference and your photos can reveal sexual orientation, race, religion or health, so we ask for your separate consent for this, which you can withdraw in Settings. If other people appear in your screenshots, they are sent too, because the AI cannot review a photo without seeing it.
What never goes to Anthropic
We do not send your account ID, your purchase information, your consent records, your device check data or any city. There is no email or name to send, because accounts are anonymous. Your screenshots themselves can identify you, because they show whatever your profile shows, such as your face, your first name or your job. We do not ask the AI to identify anyone from their face, or to guess anyone's race, religion, health or sexual orientation. We do not ask the AI to create face templates or scans and we store none.
What the AI is asked to do
One step checks safety, including whether anyone who is the subject of the profile appears to be under 18. If so, the review stops and we delete those screenshots at once. That is the only personal characteristic the AI is asked to look for. Another step judges how the profile is presented, meaning the photo choices and the writing. The AI is told never to judge or mention how attractive anyone is, their body, weight, height, skin, race, age or apparent gender. The AI also gives each profile a presentation score from 0 to 100 with a one-sentence reason. We use it to pick your band, we store it on our server until you delete your account, and you never see it.
What Anthropic does with it
Anthropic deletes API inputs and outputs within 30 days. Content flagged for trust and safety may be kept up to 2 years, and safety classifier scores up to 7 years. Anthropic says it does not train its models on API data, except that content flagged as violating its usage policy may be used to train its safety classifiers. We do not use your personal data to train large language models. See Anthropic's privacy center.
How long we keep what the AI wrote
Your screenshots are deleted after 30 days. What the AI wrote from them stays until you delete your account: the descriptions of each photo and prompt, the verdicts and editor lines, the cover headline, your rewrites, and the hidden score with its reason. There is no way to delete one issue. The Privacy Policy has the full table.
The output is AI-generated and can be wrong
Your critique, including the band (weak, okay or strong), the verdicts and the rewrites, is written by an AI, not by a person. The app marks it "Written by AI (Claude)" on the cover, the verdicts, the rewrites and the share card. It can be wrong, blunt, inconsistent or tone deaf, and the same profile may get different answers twice. A band is not a measure of your worth. It is not professional advice. It is automated feedback and is not used to make any decision with legal or similarly significant effects on you.
No human reviews it by default
No person at Fruitful LLC reads your screenshots or critiques as a matter of course. The AI's output goes straight to you. If you report a problem, we will read what you send us, and the report points us to the issue and item it is about.
How to report a problem
Use "Report this" on a critique or verdict in the app. It sends us the report without leaving the app. You can also email jylee@fruitfulfutures.com and tell us what happened. Please include your account ID from Settings, because accounts are anonymous and we cannot find your data without it.
More
See the Privacy Policy for how long data is kept and your rights, and the Terms of Service for the rules of use.
Contact
Fruitful LLC, 82 Wendell Ave Ste 100, Pittsfield, MA 01201. jylee@fruitfulfutures.com.