Norm← Back to Norm

Norm — Privacy Policy

Published at: norm.welab.love/privacy Last updated: 11 July 2026 · Version: 1.6 (review cadence in Section 6 updated to about every three weeks, matching the adopted human-review process; v1.5 same day: Section 4 quotes the real shipped consent checkbox, which now names the Art 9 special category — sex life / sexual orientation)

The short version: no account, no name, no profile. We can't tell who you are, and we built it that way on purpose. The longer version's below — it's the honest one, so it names the single place your words get stored, and what we do about it.


1. Who we are

Norm is operated by Noy Carmel, working as an autónomo (self-employed) in Spain, as part of WeLab (welab.love). Noy Carmel is the data controller for the processing described here.

Questions or requests: info@welab.love.

This policy is published at norm.welab.love/privacy.

2. How Norm is built, in one paragraph

Norm has no accounts, no login, and no profiles. Norm's chat doesn't ask your name. Our database stores no session identifier, no account ID, no IP address, and no shared key that links one record to another — so we have no straightforward way to reconstruct a chat history by person, or to trace what you typed back to you from what we store. These are facts about how the system is built, not just promises. Two honest notes: you can still type identifying details into a message yourself, and booking a consult or joining the newsletter are separate steps where you choose to give contact details (see Sections 8 and 9).

You're a stranger to us, and you stay one.

3. What we handle — and what we deliberately don't

Here's everything Norm itself collects and stores, including the boring parts. If it's not on this list, Norm doesn't keep it (your host's own server logs are separate — see the footnote below).

DataWhere it livesNotes
Your conversation (what you type + Norm's reply)Our EU databaseThe only place we store chat content. Treated as sensitive — see Section 4.
Coarse traffic info: 2-letter country, device type, date, hour of dayOur EU databaseNo content, no identifier, no precise time.
Which references Norm used + exchange number + a safety flag for the turnOur EU databaseHow Norm picked what to say, plus a one-word flag for the kind of turn it was — usually "general," sometimes a safety flag such as self-harm or minor. No content, no time, no location, and nothing that links this row to you or to your conversation.
Short-term rotation memoryOur EU databaseAuto-deleted after 30 minutes. Nothing identifying.
Newsletter email (only if you give it)A separate email providerSee Section 8. Never stored with your conversation.
Booking details (only if you book)Passed to a scheduling serviceSee Section 9. We don't store them in our own database.
Feedback (only if you use the feedback form)Our EU database, a separate tableSee Section 3a. No name, no email. Not linked to your conversation.

What we deliberately don't collect or store in our database: your name, any account or login, your IP address, any cookie, any cross-site tracker, any session ID, and any timestamp on the conversation text itself.

One honest footnote — our host's plumbing. Like every website, Norm runs on a hosting provider, and that provider keeps its own basic server logs, which can include rough technical details like IP addresses. That sits at the infrastructure layer, on their side, separate from Norm's own database.

We don't collect that, don't want it, and don't go digging in it — and it never touches what you actually said to Norm. It's standard plumbing every host has; it's their record, not ours.

3a. Feedback (separate, and entirely optional)

Spotted something weird, or have an idea? There's a small "tell the humans" link in the chat. It's entirely optional, and we don't ask for anything that could identify you.

Sending feedback is voluntary — we ask for no name and no email. What you send (a category, your own words, and up to two optional screenshots) is stored in our EU database, in a table kept separate from the chat logs in Section 3 and not linked to your conversation: no shared identifier, no shared timestamp convention, no way to join the two. One honest note: a screenshot can show your chat on screen, so only send what you're comfortable with a real human reading. Feedback is deleted within 45 days, on the same schedule as everything else in Section 6.

4. Why your conversation is treated as sensitive

People talk to Norm about desire, orientation, and sexual difficulty. Under EU law that's special-category data (Article 9 GDPR). We treat it that way even though we can't tie it to you — which is why we ask you to agree before any of it is stored.

The entry moment (the same screen as the age check). Before you come in, one screen asks you to confirm you're 18 or older, and to tick a box agreeing that parts of your chat are stored so a real human at WeLab can check Norm is doing right by people. The box you tick says exactly this:

"Norm and WeLab can save bits of my chat — including what I say about my sex life and sexual orientation — minus my name, which I wouldn't share anyway. A real human at WeLab reads a sample, so Norm keeps doing right by people. Wiped within 45 days."

Before any conversation text is written to our database, an automated step strips names that refer to you or to people in your personal life (public figures are kept). This is automatic pseudonymisation; we describe it as "we automatically strip names," not a guarantee that zero names ever slip through.

Lawful basis (per purpose):

  • Storing and reviewing your conversation → Article 6(1)(a) consent + Article 9(2)(a) explicit consent, taken at entry.
  • Content-free traffic, retrieval, and rotation data → Article 6(1)(f) legitimate interest (running and improving the service safely; no content and no identifiers are involved, so the impact on you is minimal).
  • Newsletter → separate consent (Section 8). Booking → to provide the booking you asked for (Section 9).

5. Why we keep anything at all

To produce Norm's reply (your message goes to our AI provider — Section 7); to rotate and de-duplicate Norm's references so it doesn't repeat itself; to understand at a coarse level where and when Norm is used; and so a human at WeLab can read a sample of real conversations to check Norm is responding safely and well (this is the named safety check that gates public launch — and the reason conversation text is stored at all).

6. How long we keep it

  • Short-term rotation memory: 30 minutes, auto-deleted.
  • Coarse traffic and reference data: deleted at least every 45 days.
  • Your conversation: deleted at least every 45 days. A human at WeLab reviews a sample regularly (currently about every three weeks) to keep Norm safe, and the logging tables are wiped on a monthly cycle, always within 45 days.
  • Feedback you send us (Section 3a): deleted within 45 days, on its own daily cleanup schedule.

We don't keep this forever. At most 45 days, a human glances at a sample to make sure Norm's behaving, then it's gone.

Retention is enforced by a periodic full wipe of the logging tables rather than by per-record dating, which keeps the no-timestamp design intact.

7. Who else touches your data

To run Norm we use a short, fixed set of providers. The only one outside the EU is our AI provider, Anthropic (USA), which generates Norm's replies:

WhoRoleWhereTransfer safeguard
Anthropic (Claude)Writes Norm's replies; your message is sent for processingUSAStandard Contractual Clauses (EU 2021/914, Modules 2 & 3), via Anthropic's data-processing agreement
Our database providerStores the data in Section 3EUWithin EU
Our email providerNewsletter onlyEUWithin EU
Our scheduling service (Cal.eu)Booking onlyEU (Cal.eu data residency)Within EU — processor under a signed Cal.com Data Processing Agreement

About the AI step: every message you send is processed by Anthropic's Claude to write a reply (a short series of passes to draft and quality-check it, plus an automated redaction pass on stored exchanges that strips personal names before storage), so your message leaves our EU database boundary at that moment. Anthropic acts as our processor and does not train its models on this data under our agreement.

Anthropic is the only non-EU recipient of your chat content; the booking path stays in the EU on Cal.eu, and the newsletter stays with an EU provider. (Our hosting provider's own technical logs sit at the infrastructure layer and are separate from this — see Section 3.)

Norm's brain is Claude, made by Anthropic — so when you hit send, your words go there to get a reply, then come back. That's the one trip your chat content takes outside Europe, and it's covered by a formal data agreement.

The full named list of every provider (with regions) lives in our internal Record of Processing Activities. Cal.com acts as our processor for booking (we set the event type and purposes), governed by a signed Cal.com Data Processing Agreement.

8. Newsletter (entirely separate)

If — and only if — you choose to give your email to keep going, it goes to our email provider and nowhere else. It is never linked to, or stored with, your conversation — the two systems share no key, so they can't be joined. Single opt-in; unsubscribe anytime.

Your email goes only to the newsletter list (the provider that delivers the emails), it's never linked to your conversation with Norm, and it's never handed to anyone else for their own use.

9. Booking a consult

If you book a free consult, the details you enter (name, email, phone, timezone, and the goal you describe) go straight to our scheduling service to make the booking. Norm doesn't store them in our own database. And if you leave the "share my goal" toggle off, the goal you typed is not sent at all — the booking carries only a generic placeholder.

10. Your rights, honestly

Because we hold no name, account, or identifier, we usually can't find "your" data even if you ask — there's no handle to look you up by. This is Article 11 GDPR: we process data that doesn't let us identify you, so some of your rights can be limited where we genuinely cannot single you out, and we won't collect extra identifying details just to try. That's a consequence of the privacy design, not a dodge.

So: by default we can't action an access or deletion request. But if you give us enough to locate a specific exchange — a verbatim snippet — and finding it is feasible, we'll delete it. You can always unsubscribe from the newsletter, reach us at info@welab.love, and lodge a complaint with Spain's data protection authority (AEPD, aepd.es).

Weird thing about being anonymous: we genuinely can't pull up your chat to delete it, because we don't know which one is yours. If you paste us the exact line, we'll go find it.

11. You have to be 18

Norm is for adults. You confirm you're 18 or older before chatting; if you're under 18, Norm points you to resources made for your age (Scarleteen, Amaze) instead. This 18+ confirmation is the same entry moment that carries the data-storage consent in Section 4.

12. Cookies / local storage

Norm uses your browser's temporary session storage only to remember basic things during your visit (like passing the age gate). Norm itself doesn't use advertising or analytics cookies, or third-party trackers — so there's no cookie banner. Our fonts are served from our own domain, not fetched from a third party as you browse.

13. Changes, and the fact that Norm is an AI

We date and version this page and note material changes. Norm is a bot — you're talking to software, not a person.

Norm's also just a bot, so he gets things wrong sometimes. Then again, so do humans.


Norm — operated by Noy Carmel / WeLab, Spain · info@welab.love

← Back to Norm