New: Search Now Knows What a Style Is For

Read →

Privacy

Last updated 15 November 2026 · draft

The short version

  • We use your data to run Midlibrary. We never sell it, and there are no ads.
  • Google Analytics counts visits. In the EU, the EEA, the UK, Switzerland and California it runs only if you say yes. Your answer ↓
  • With an account, we keep your email, your profile, the styles you save and your membership — and what you view and copy, if your activity switch is on.
  • Your profile is public. Each collection is public, members-only or private, as you choose.
  • You can see, fix, copy or delete your data, any time.

Who we are

Midlibrary is run by JB Nina, Inc., 131 Continental Drive, Suite 305, Newark, Delaware 19713, USA. We decide how your data is used; in law, we are the “controller”. Write to us at librarian@midlibrary.io.

What we collect, and why

Everyone who visits

  • Connection data — when your browser asks for a page, Cloudflare receives your IP address, your browser type and the page you want. Cloudflare uses them to deliver the page and to protect the site from attacks.
  • Your country (in the US, your state) — read from that connection data, only to decide whether to ask you first: about Google Analytics, style counts and your activity. We don’t store it.
  • Page speed — Cloudflare Web Analytics measures how fast our pages load and counts visits. It uses no cookies and doesn’t recognize you on a later visit.
  • Google Analytics, if allowed — see Analytics — your choice.
  • Form checks — Cloudflare Turnstile checks that a log-in, sign-up or message comes from a person. It looks at signals from your browser. Usually you see nothing.
  • Log-in protection — we count log-in attempts per IP address and per email address, to stop people guessing codes. The counts delete themselves within an hour.

If you make an account

  • Your email — to send log-in codes, notices about your account and membership, and replies.
  • Your name, picture, About text and profile address — to show your public profile. Anyone can see it.
  • Your collections and the styles you save in them — to keep your library. You set each collection to public, members-only or private.
  • Your likes — which styles you like, kept with your account. Others see only each style’s count, never who liked it.
  • Your notes (members) — the notes you write on a saved style or on a collection. Only you see them. A style’s note goes when you remove it from that collection, and all of them go when you delete your account.
  • Your display preferences — to show the site the way you chose.
  • Which activity blocks you show on your public profile (stats, achievements, your look, liked styles) — all off until you turn one on.
  • Your answers to “Help us know you” (four optional questions) — to understand who uses Midlibrary. Only we see them.
  • Your newsletter choice — to respect it.
  • Dates, internal numbers and your Stripe customer number — to run your account and membership.
  • The style you joined from — if you made your account from a style’s page, which style, so we know what brings people to Midlibrary.

We also keep which styles you view and copy, if “Keep what I view, copy and save on my account” is on for you (see below).

If you log in with Discord, Patreon or Google, that service shares your name, email and picture with us, and we store them. If you support us on Patreon, Patreon tells us your tier, so we can give you the matching membership.

You need an email address to have an account. Everything else is optional.

If you buy a membership

We keep your plan, its status and its dates. Stripe takes the payment and keeps your card and billing details. We never see your card number. We keep a log of the payment notices Stripe sends us, by their reference number.

When you pay by card, we keep a record of what you agreed to: the renewal terms as they were shown, the price and the date. If you withdraw within 14 days, we keep a record of the withdrawal and the refund.

When you open the membership offer, we note which button opened it, which tier it showed and whether you went on to pay — with no name, email or IP address, so it says nothing about who you are. We use it to see what makes the offer work.

If you pay by card, your name and picture are shown in the supporters’ hall under your tier. You can hide them in Account.

If you write to us or report a bug

We keep your message, your name and email (if you give them or are logged in), the page’s address, your browser’s user-agent (a line naming your browser and system) and your screen size. A copy of each message goes to our team’s inbox. If you add a screenshot, it goes only to that inbox; the site doesn’t keep it. If you delete your account, the messages you sent stay, without your name and email.

If you answer a poll

We sometimes run a short poll on the site. Your answers are kept without your account: we note only whether you were signed in, whether you are a member and your plan. Your browser keeps a random code so you can change your answers until the poll closes; we keep only a scrambled copy of it. What you write in a text box may be read by our team and sorted into themes with the help of an AI service. The results are for us, never published with anything that identifies you.

If you join the newsletter (when it starts)

Our newsletter is starting soon. This part applies from the first issue.

  • It comes from “Andrei from Midlibrary” and is sent by Resend from news.midlibrary.io.
  • We keep your email address and your choice. Resend keeps a copy to send the emails.
  • If you sign up in the footer, we first email you to confirm. You join only after you click.
  • We never email members to ask them to subscribe. Members see a one-click invitation on the site instead.
  • Every issue has a one-click unsubscribe. We stop right away. We keep a note that you left, so we never email you again by mistake.

Style counts

We count views, copies, saves and searches of each style, and clicks on a style in a grid with its place there, so the Library can show what is popular. These counts use no cookies and store nothing on your device. To tell a repeat view on the same day, our server makes a code from your connection details and scrambles it with a key that changes every day. It keeps that scrambled code only until the day ends, when the code and the key are both deleted, and the counts never carry it. We keep each count for 3 months without anything that identifies you, and the daily totals for good. When you copy a style’s prompt or SREF code, our server sends you the text and counts that copy, wherever you are: like a view, the count holds nothing that identifies you. In the EU, the EEA, the UK, Switzerland and California, we count your clicks, saves and searches only if you said yes to analytics. Searches are never linked to you.

Your activity on your account

If “Keep what I view, copy and save on my account” is on, we keep which styles you view, copy, save and remove, where on the site, and when. We use it so your library can remember what you looked at and suggest styles you’ll like. Only you and Midlibrary see it. In the EU, the EEA, the UK, Switzerland and California it stays off until you turn it on. Elsewhere it is on, and you can turn it off. Turning it off erases what we kept and keeps nothing new. Deleting your account erases it too. We never keep your searches on your account.

From it we work out, each time you open them, your stats, your achievements, the styles “For you”, “Your look” (the looks and colours you come back to) and, for members, your year in styles. We don’t store any of them: they come from what we keep above, so turning your activity off or deleting your account removes them too. A look you mute is kept until you unmute it, turn your activity off or delete your account. Only you see all this, unless you turn a block on for your public profile: then anyone can see that block, and only it, until you turn it off. Muted looks are never shown there.

Artists named in the Library

Our style pages name artists as style references. For a living artist, the name and a link to the artist’s own website are personal data. We take them from public sources, such as Wikidata, and check them ourselves. We use them to teach about visual styles, on the basis of our legitimate interest. Anyone can see these pages, and search engines show them. We show no works by any artist.

We did not write to each artist, because that would take an effort out of proportion to this limited use; this notice takes its place. An artist, their estate or their representative can object at any time: we remove the name, the page and the links within 30 days of a verified request (see For artists). You can also complain to a data protection authority.

If you had an account on our old site

On 21 October 2026 we moved Midlibrary to new technology. We carried over accounts, collections and memberships. Collections that were public on the old site are public now too. You can make any collection private or members-only on the Collections tab.

The old newsletter list stays in an archive at Airtable and SendGrid only until the newsletter moves to Resend.

Analytics — your choice

Google Analytics tells us how people find Midlibrary and whether they come back. It runs through Cloudflare’s servers, so your browser loads no Google script. It does set cookies to recognize your next visit (see Cookies).

It also tells us when someone clicks an artist’s link (which style page, and which kind of link) and when someone copies a style’s code or prompt (which style).

We have turned off Google’s advertising features and Google signals. Google keeps visit-level data for 2 months, the shortest it allows. Totals stay in our reports.

In the EU, the EEA, the UK, Switzerland and California, it runs only if you say yes. If we can’t tell where you are, we ask. Anywhere else, it runs unless you turn it off. You can change your answer here at any time. Saying no changes nothing else about the site.

If your browser sends a Global Privacy Control signal, we treat it as “Off” for Google Analytics, unless you turn it on below. Browsers’ older “Do Not Track” signal has no common standard, and we don’t act on it; use the switch below instead. Apart from Google Analytics as described here, we let no other company collect data on Midlibrary to follow you across other sites.

Google Analytics may count my visits

On — outside Europe it counts unless you turn it off.

Cookies

We use no advertising or cross-site tracking cookies.

  • ml4-session — keeps you logged in. Needed. 30 days.
  • ml4-oauth — carries a log-in through Discord, Patreon or Google and back. Needed. 10 minutes.
  • ml4-signup — holds a sign-up in progress. Needed. 30 minutes.
  • ml4-analytics — remembers your analytics answer and its date. Needed for your choice. 12 months, then we ask again.
  • cf_consent — passes your analytics answer to Cloudflare’s analytics service. Needed for your choice. 1 year.
  • cf_clearance — shows Cloudflare that this browser passed a security check, so the site stays safe. Needed for security. Up to 1 year.
  • ml4-settings — remembers your theme and page size when you’re logged out. Your preferences. 12 months.
  • ml4-help-doors — remembers that you closed the help hints. Your preferences. 12 months.
  • ml4-poll — remembers that you answered the poll or asked not to see it. Your preferences. 30 days.
  • ml4-poll-token — lets you change your poll answers from the same browser. Needed for your answers. 30 days.
  • ml4-nudity-shown — remembers that you said you are 18 or older, so pictures with nudity show without their veil. Kept in your browser’s storage, not a cookie, and never sent to us. Your preferences. Until you clear this site’s data.
  • ml-help-side — remembers whether you last read Help for Midjourney or for ChatGPT. Kept in your browser’s storage, not a cookie, and never sent to us. Your preferences. Until you clear this site’s data.
  • ml4:similar-path — remembers the styles shown in this tab, so Similar styles doesn’t show them again. Kept in this tab’s storage, not a cookie, and never sent to us. Needed for the feature. Until you close the tab.
  • ml4-news — remembers that you closed or read the newest news post, so its bar stays away. Your preferences. 30 days.
  • cfz_google-analytics_v4, cf_zaraz_client — Google Analytics, through Cloudflare: recognizes your browser on a later visit. Analytics, only while allowed. 1 year.
  • cfzs_google-analytics_v4 — Google Analytics, through Cloudflare: one visit. Analytics, only while allowed. Until you close the browser.

The needed and preference cookies don’t need your consent: they only remember what you chose or keep the site safe. You can delete cookies in your browser at any time. Then the site forgets your display choices and hints, and may run its security check again.

Who handles your data for us

These companies work for us. They may use your data only to provide their service to us.

  • Cloudflare — hosting, database, file storage, security, Turnstile and page-speed measurement; runs Google Analytics on its servers. USA and a worldwide network. EU-US Data Privacy Framework and standard contractual clauses.
  • Bunny.net — stores and delivers pictures, profile pictures included. Stored in Germany (EU), delivered from servers worldwide. An EU company.
  • Resend — log-in codes, account emails, copies of your messages to our inbox, and the newsletter (when it starts). USA. Standard contractual clauses and the EU-US Data Privacy Framework (with its UK extension).
  • Stripe — membership payments and billing records. USA and Ireland. EU-US Data Privacy Framework and standard contractual clauses.
  • Google — Google Analytics, only as described above. USA. EU-US Data Privacy Framework and standard contractual clauses.
  • Zoho — our team’s inbox (librarian@, community@): your messages and screenshots. Its EU data center.
  • Airtable, SendGrid — the old newsletter list, as an archive only. USA. Standard contractual clauses.

These companies act for themselves, under their own privacy policies, when you choose to use them: Discord, Patreon and Google for log-in, and Patreon for support. Stripe also keeps some data for its own fraud checks and legal duties.

We post pictures to Midlibrary’s own Pinterest and Instagram accounts through Pinterest’s and Meta’s APIs. Each app signs in only to our own account; we receive and keep no data about other Pinterest or Instagram users.

We also use AI assistants from Anthropic and OpenAI to help run Midlibrary. They may see messages and account data, and handle it under their own privacy terms (Anthropic, OpenAI). We’ve turned off their use of it for training.

Some of these companies work outside your country, including in the USA. Where EU, UK or Swiss law applies, we use the safeguard named above. Write to us for a copy.

How long we keep it

  • Account, profile, collections, saved styles, answers — until you delete your account. Deleting removes them at once, and ends any card membership. Our backups forget them within 30 days.
  • A deleted account’s internal number and the date — for good, so a copy from the old site can never bring the account back.
  • Membership plan and dates — while you are a member, then as billing records.
  • Billing records — as long as tax law requires.
  • Your agreement to automatic renewal, and any withdrawal and refund — three years, even if you delete your account: the law asks us to be able to show them. They hold your account’s internal number and Stripe’s references, not your name or email.
  • Messages and bug reports — 12 months after you send them, then deleted, from the site and from our inbox.
  • Log-in attempt counts — up to an hour, then they delete themselves.
  • Google Analytics — visit-level data for 2 months. Totals remain.
  • Newsletter (when it starts) — until you unsubscribe. Then only a note that you left.
  • Style counts — single counts, with nothing that identifies you: 3 months. Daily totals: kept.
  • Your activity on your account — until you turn it off or delete your account. The scrambled code for repeat views — until the day ends.

Your rights

You can see the data we hold about you and get a copy, correct it, delete it, ask us to pause using it while we check something, and take back any consent you gave.

Your right to object. Where we rely on our legitimate interest, you can object at any time, for reasons about your situation. We will stop unless we have strong legitimate reasons to continue. You can always object to marketing emails, and we will stop.

Most of this you can do yourself: in your profile you can edit everything or delete your account. For a copy of your data, or anything else, write to librarian@midlibrary.io. We answer within one month. If a request is complex, we may need up to two more months; if so, we’ll tell you why within the first month. We may ask you to write from your account’s email, so we know it’s you.

If you think we got it wrong, you can complain to a data protection authority, for example where you live or work. In the UK that is the Information Commissioner’s Office. In Switzerland it is the Federal Data Protection and Information Commissioner.

Children

Midlibrary is for adults. You must be 18 or older to make an account or buy a membership, as the Terms of Use say. If you think a child has given us their details, write to us and we’ll delete them.

Changes to this page

When we change this page, we change the date at the top. If a change matters to you, for example a new use of your data or a new company handling it, we email members at least 30 days before it takes effect. If a change needs your consent, we ask for it first.

Questions

librarian@midlibrary.io · JB Nina, Inc., 131 Continental Drive, Suite 305, Newark, Delaware 19713, USA