Biometric Data Policy
Effective November 13, 2026 · Version 2.7
Kind’s photo verification briefly processes a measurement of facial geometry to check that a live capture matches a reference photo, and to check new profile photos against the faces of banned accounts. This is our written retention and destruction policy for that data, published as required by the biometric privacy laws of Illinois, Texas, and Washington.
The short version: if you pass, your verification photos are deleted within seconds and only the result is kept; if you do not, we keep one pair privately, a person sees it only if you ask for a review, and it is deleted on a fixed schedule; we store a face template only for an account we ban, for no longer than one year; we never sell or disclose biometric data to anyone; and we ask for your agreement before the camera opens.
1. Why This Policy Exists
Kind, operated by Subtxt LLC, processes a measurement of facial geometry in two places: a photo verification check, and a check of new profile photos against the faces of banned accounts. Several laws — including the Illinois Biometric Information Privacy Act (740 ILCS 14), the Texas Capture or Use of Biometric Identifier Act (Tex. Bus. & Com. Code § 503.001), and Washington RCW 19.375 — require a company that does this to publish a written policy setting out how long it keeps biometric information and how it destroys it. This is that policy. It forms part of our Terms of Service and our Privacy Policy.
2. What We Collect and How Verification Works
Verification is required to join
Every Kind member verifies. That is what makes the badge worth anything, and it is one of the main defences against catfishing on a platform where members are sharing health information. If you do not want facial geometry processed, do not complete verification — you will not be able to finish creating an account, and you can email us to confirm that anything already entered has been deleted. You are never charged before verification, and no profile is created until it succeeds.
When you verify, this is exactly what happens:
- You choose a reference photo of your face: one of your profile photos, or a private photo from your device that is not published on your profile.
- You take a live capture with the front camera, inside the app. The capture is three frames: facing the camera (the "live selfie"), turned left, and turned right. We check the movement between the frames for signs that a live person is present.
- All images are sent to our server over an encrypted connection and processed in server memory.
- Face-detection software locates a face in each image. If it cannot find one, verification fails immediately and nothing further happens.
- Each detected face is converted into a numeric vector describing facial geometry — a "face template". The template is a list of numbers. It is not a picture and cannot be turned back into one.
- The two templates are compared to produce a single similarity score.
- The score is compared against a threshold to produce a pass or fail result.
- Both templates and the turned frames are then deleted from memory. If the check passes, the reference photo and the live selfie are deleted too, within seconds, and only the result, the scores, and a time stamp are saved. If the check cannot match them, the live selfie and the reference photo are kept for a possible human review, as set out in Section 3.
The biometric identifier we process is a scan of face geometry, and the biometric information we derive from it is the face template. These are the only categories of biometric data Kind handles. We do not collect fingerprints, voiceprints, retina or iris scans, hand geometry, or DNA, and we do not collect face geometry for any purpose other than the two described in this policy. Voice notes are recordings you publish on your own profile; we do not analyse them to identify anyone and we do not create voiceprints.
The banned-account face check
A person we ban can come back with a new phone number and new photos. To find them, we check new profile photos against the faces of accounts we have banned. The face check applies only to members who have agreed to it, when they sign up or when they verify again. Other members’ new photos get only the image check in step 1 below, which is not biometric.
- When you save a profile photo, our server compares it with the photos of banned accounts using an image fingerprint (a perceptual hash). This finds copies of a banned photo. The fingerprint describes the image file, not your face, and it is not biometric data.
- If that does not settle it, face-detection software locates the largest face in the photo and converts it into a face template in server memory.
- That template is compared with the stored templates of banned accounts and is then deleted from memory. It is not stored.
- If the faces match, the account can be limited until a person at Kind reviews it. The software does not ban anyone. A person makes that decision.
- If we ban an account that this check applies to, we make a face template from each of its public and private album photos that shows a face, and we store it as set out in Section 3. We use a stored template only to recognize the same person on another account.
3. Retention Schedule
Face templates are stored only for banned accounts, and we keep at most one verification photo pair
A face template made during verification, or when we check a new profile photo, exists only in volatile server memory for the duration of a single comparison — a few seconds — and is destroyed as soon as it completes. It is never written to disk, to a database, to object storage, to a log, or to a backup. The only face templates we store belong to accounts we have banned, as set out below. If a verification check passes, the reference photo and every live capture frame are destroyed in the same way, within seconds. If the check cannot match the two faces, Kind keeps one pair of images — your most recent live selfie and the reference photo it was compared with — in private, access-controlled storage, so that you can ask a person to check the result. A newer failed attempt replaces the older pair, so we never hold more than one pair for any person.
No employee, contractor, or reviewer sees a retained pair unless you ask for a review. If you do, one trained reviewer compares the two images, once, to decide your verification. A retained pair is used for nothing else.
A retained pair is permanently deleted at the earliest of: (a) the moment you pass verification; (b) immediately after the reviewer decides your verification; (c) when you delete your account; or (d) 30 days after your last failed attempt, if no review is pending.
| Item | Where it lives | Retention |
|---|---|---|
| Reference photo | Server memory; private storage only after a failed check | Check passed: destroyed within seconds. Check failed: kept as the retained pair until deleted as set out above |
| Live selfie (front-facing frame) | Server memory; private storage only after a failed check | Check passed: destroyed within seconds. Check failed: kept as the retained pair until deleted as set out above |
| Turned-left and turned-right frames | Server memory only | Destroyed immediately after the comparison |
| Face templates from verification (both) | Server memory only | Destroyed immediately after the comparison |
| Face template from a new profile photo | Server memory only | Destroyed immediately after the comparison |
| Face templates of a banned account | Private database table | Until the ban is lifted or the account is permanently deleted, and no longer than one year after the ban |
| Image fingerprints of a banned account’s photos (not biometric) | Private database table | Until the ban is lifted or the account is permanently deleted |
| Similarity score, pass/fail result, time stamp | Database | Life of the account, then up to 12 months as an anti-fraud record |
| Verified badge status | Database | Life of the account |
The similarity score and result are not biometric identifiers or biometric information. They are a numeric outcome from which no facial geometry can be reconstructed. We keep them so that we can show your verified badge, detect repeated fraudulent verification attempts, and answer support questions.
Stored templates of banned accounts
We store face templates only for accounts we have banned: one for each photo that shows a face, in a private database table that members cannot reach. Every stored template is permanently deleted at the earliest of: (a) the moment the ban is lifted; (b) when the account is permanently deleted; or (c) one year after the ban.
For the avoidance of any doubt: we permanently destroy any biometric identifier or biometric information we hold, and any retained photo pair, at the earliest of (a) the moment the purpose for collecting it is satisfied, (b) the deletion events set out above, or (c) one year after your last interaction with Kind — well inside the three-year outer limit set by the Illinois Biometric Information Privacy Act — and in any event promptly upon your request.
4. Destruction Guidelines
- Image bytes and template vectors are held in local variables scoped to a single request and are explicitly released when the comparison ends, including when it ends in an error.
- The verification code path writes no template to disk, to object storage, to a database column, or to any cache. The only images it writes are the retained pair after a failed check, to private storage reserved for that purpose, and each failed attempt overwrites the previous pair.
- A retained photo pair is deleted from storage on the events set out in Section 3.
- The check of a new profile photo writes no template anywhere. A template is stored only when we ban an account, as set out in Section 3. It is deleted at once when the ban is lifted, and a nightly job deletes any that has reached its limit.
- Application logs record only the outcome, the score, and non-identifying diagnostics. Image bytes and template values are never logged.
- A template made during verification or a photo check is never persisted, so it cannot enter a backup or a replica. A stored template of a banned account is in our database, so a copy can remain in a database backup until that backup expires on our provider’s rolling schedule.
- These controls are part of the verification service itself, so they apply on every attempt, including failed and abandoned ones.
- If a future change to Kind ever required storing any other face templates, or keeping any other biometric data beyond what this policy describes, we would first update this policy, publish the new retention and destruction schedule, and obtain fresh written consent before collecting anything.
5. Notice and Consent
Before any capture, the app tells you in plain language what verification does with your photos, that facial geometry is measured, that your photos are deleted within seconds if you pass, and that one pair is kept for up to 30 days if you do not. It then asks you to agree, and you must take that affirmative step before the camera opens. That agreement is your written release. We record that you gave it, the date and time, the version of this policy you saw, and the app build you saw it on, and our server refuses to process any capture without it. You agree to the banned-account face check through the Terms you accept when you sign up, which include this policy, and current versions of the app also tell you about it on this screen.
You may withdraw consent at any time by emailing us. Withdrawal takes effect immediately, and the banned-account face check stops for your account. Verification templates are never stored, so there are none to delete, and if we hold a retained photo pair or a stored template for you we will delete it promptly. Withdrawing does not delete your account, and it does not remove an existing verified badge unless you ask us to remove it — but you will not be able to re-verify until you consent again.
6. We Never Sell or Disclose Biometric Data
Absolute commitments
Kind does not and will not sell, lease, trade, or otherwise profit from any biometric identifier or biometric information. Kind does not disclose, redisclose, or otherwise disseminate biometric data to any third party — not to advertisers, not to data brokers, not to affiliates, and not to service providers, other than the infrastructure providers that host our private storage and our database, which hold a retained photo pair or a stored template only on our behalf, under contract, and cannot use them for anything else.
The face-detection and face-embedding models run on our own servers. Your images are never sent to a third-party face recognition service, and the model providers receive nothing from us. The only circumstances in which biometric data could ever be disclosed are those the law does not let us refuse — a valid warrant or subpoena, or a court order — and we store no face template except for a banned account. At most, we could hold the single retained photo pair and, for a banned account, the stored templates described in Section 3, for the limited time set out there.
7. Safeguards
We protect biometric data using the reasonable standard of care for our industry, and at least as protectively as we protect other confidential and sensitive information. Images travel only over encrypted HTTPS/TLS connections. Processing happens in an isolated server process. Face templates are stored only for banned accounts, in a private database table that members cannot reach. A retained photo pair is kept in private, access-controlled storage and is opened only to handle a review you ask for. Access to the production environment requires multi-factor authentication and is logged.
8. Rights by State
Illinois
If you are an Illinois resident, the Biometric Information Privacy Act gives you the right to be told in writing that biometric data is being collected and why, to be told how long it will be kept, and to give or withhold written consent before collection. This policy and the in-app consent screen provide that notice. You may withdraw consent, and you may ask us to confirm what biometric data we hold about you — we hold no face template unless we have banned the account, and at most we hold the single retained photo pair and those templates, as described in Section 3.
Texas
If you are a Texas resident, we inform you before capturing a biometric identifier and obtain your consent, we do not sell or disclose it, and we destroy it within a reasonable time and in any event within one year of the purpose ending — in practice, within seconds of a passing check, and otherwise on the schedule in Section 3.
Washington
If you are a Washington resident, we give notice and obtain consent before enrolling any biometric identifier in a database for a commercial purpose. We enrol a biometric identifier in a database only for a banned account, to stop that person returning, as set out in Sections 2 and 3. We do not sell, lease, or disclose biometric identifiers. See also our Consumer Health Data Privacy Policy.
Other states
Biometric data is "sensitive data" or "sensitive personal information" under the privacy laws of California, Colorado, Connecticut, Virginia, and other states. We process it only with your consent and only for the verification and banned-account purposes described here, and you may exercise the rights set out in our Privacy Policy.
9. What Verification Does Not Mean
A verified badge is not an identity guarantee
Verification is a best-effort signal — automated, or decided by a reviewer if you ask for a review — that a live capture plausibly resembles a reference photo. It is not identity verification, an age check, a background check, a criminal records search, or a sex offender registry search, and it is not a warranty of anyone’s identity, honesty, health status, or safety. Automated matching produces false positives and false negatives. Never let a badge substitute for your own judgement — see the safety guidance in Section 6 of our Terms of Service.
10. Questions and Requests
Email info@kind.date with "Biometric Policy" in the subject line. Subtxt LLC is a Wyoming limited liability company operating remotely, so email is how we receive these requests. We will respond within 45 days.
We will post any change to this policy here with a new effective date and version number, and we will give notice in the app before a material change takes effect.
Version 2.3 (October 30, 2026): Sections 2 to 9 now describe the one photo pair we keep after a failed check, who can see it, and when it is deleted.
Version 2.4 (October 31, 2026): No change to this policy. The version number follows the Consumer Health Data Privacy Policy, which now explains how to change or remove your HSV type.
Version 2.5 (November 4, 2026): Sections 1 to 8 now describe the check of new profile photos against the faces of banned accounts, and the face templates we store for banned accounts.
Version 2.6 (November 4, 2026): Shorter wording in Sections 2, 3, and 5. No change to what we do.
Version 2.7 (November 13, 2026): Section 5 now lists what the shorter in-app consent screen says. No change to what we do.