Privacy Policy
Effective: November 11, 2026
This policy explains what Iminify collects when you use https://www.iminify.com, why, how long we keep it, who else sees it, and what you can do about it. It applies whether you use the tool as a guest or with an account.
The short version
- Your images are compressed on our own servers in the European Union. No other company receives them to compress them.
- AI alt text is on unless you switch it off: a small copy of each image (at most 768 pixels on its longest side) goes to Anthropic, whose model describes it. Anthropic deletes it within 30 days and doesn't train on it.
- We keep an upload and its optimized copy until you delete it. Nothing removes them on a timer, unless an image or upload group sent through the API asked us to (
ephemeral): then it goes about a day after it finished. An upload group whose image never arrived goes about an hour after it was opened. - An account holds your name, your email address, and a password hash or the Google, GitHub, Microsoft or Apple account you sign in with. A phone number and a profile photo are optional.
- We email you about your account, and send tips and reports on your images, which you can stop from any of them or in Settings, Emails. No email tells us whether you opened it.
- Pro and Max are billed by Lemon Squeezy. We keep the card brand and its last four digits, never the card number.
- Google Ads and Google Analytics cookies are set only with your consent if you're in the EEA, the UK or Switzerland, and can be switched off from the footer anywhere else.
- If you arrive through an affiliate's link, a Lemon Squeezy script records that visit so a purchase can be credited to the affiliate. It doesn't wait for the cookie banner.
- If you connect an AI app to Iminify, it receives what it asks for (your images' details and links, the pages you scan, your usage) under its own privacy policy. You can disconnect it at any time.
- If you connect a WordPress site, its plugin sends us the site's images, its address and name, how many images it still has to send, and the addresses of the site's pages you audit, and its key works for that site alone. You can disconnect it at any time.
- On a paid plan you can have your images' and scans' details sent to a server of your choosing when they finish. You choose the address, and you can remove it at any time.
- We don't sell personal data, and we don't use your images for anything except the compression you asked for.
Who is responsible for your data
Iminify is run by Mozex Labs, which is the data controller for everything described here. Write to [email protected] for anything about your data: questions, requests, complaints.
What we collect and why
When you visit the site
Every request to the site carries your IP address, your browser's identification string, the page you asked for and the time. We use these to serve the pages, to protect the site against abuse, and to apply the daily limits, which are counted per IP address for guests. Cloudflare, which sits in front of the site, derives your country from the IP address and passes it to us; we use it to decide whether to show you the cookie banner and to preselect your country in the phone number field in Settings, and we don't store it.
Cloudflare also measures page views and page load times for us through Cloudflare Web Analytics. It sets no cookie and builds no profile; the numbers are aggregated. Cloudflare doesn't run that measurement for visitors it serves from its data centres in the European Economic Area, the United Kingdom or Switzerland.
When something breaks, and for a small share of requests so we can see how fast the site is, the details of that request go to our monitoring provider, Flare: the address, your IP address and browser, the account you're signed in with, and the form fields you sent. Passwords, sign-in and reset codes, access tokens, cookies and session contents are left out. Our application's log entries go there too, each with the full address of the request that wrote it.
Legal basis: our legitimate interest in running and protecting the Service. Retention: application logs are kept for 14 days on our servers, the daily counters for 24 hours, and error reports, timing samples and log entries for 30 days at Flare.
When you compress an image as a guest
We store the image you upload, its name, size, dimensions and format, the options you chose, the optimized copy, and a link between the result and your browser session so the results table can show it to you. If you rename a result, we store the new name next to the original one and use it for the download. On a paid plan, when you upload a folder or a zip, we also store the path of the folder each image sat in inside it (the names of those folders), so a zip of the results puts each file back where it was. A zip is opened in your browser, which sends us the images in it one by one: the zip itself never reaches us. On a paid plan, when you ask for a responsive set, we also store the image again at each width you chose, beside the optimized copy, and delete those files with it. If you compress a result again with different settings, we store a copy of the original as a new result, so deleting either result leaves the other in place. Your IP address is used to count your daily uploads and AI alt texts, and a result compressed again counts as one upload.
Legal basis: providing the service you asked for. Retention: see "Your images" below.
When AI writes your alt text
AI alt text is switched on in the settings above the upload box, and you can switch it off before you upload. It's on for images sent through the API and through an AI app you've connected as well, unless the request sets alt_text to false. An upload group sent through the API gets one only when its request asks for it, and a connected WordPress site's images get one only when you turn it on in the plugin's settings, and only where the image has no alt text yet: as it's uploaded, or when you ask for it for images already in your library, for which the plugin sends us a small copy of each. While it's on, each image you compress gets a short description for its alt attribute and a suggested file name. We make a small copy of the image for this: at most 768 pixels on its longest side, re-encoded, with its metadata removed. That copy, and the language you picked, go to Anthropic's API, where the Claude Haiku model writes the description and the name. We store both with the result, with the language, and with a fingerprint of the original file (a SHA-256 hash) so an image we've already described isn't sent again. The small copy is deleted from our server as soon as the answer arrives, or after a day if something goes wrong. An alt text written for an image already in a WordPress library is kept with its language and a fingerprint of the small copy for a day, for the plugin to collect, and then deleted. A suggested name only renames your download if you turned on "Rename files", click "Use this name", or the request asked for it.
Anthropic processes the copy for us under its data processing agreement. It deletes what it receives and returns within 30 days, keeps it longer only if its safety systems flag it as a breach of its usage policy (then for up to 2 years) or when the law requires, and its commercial terms forbid it to train models on it. A page scan sends Anthropic a small copy only of the images the page shows without any alt text, for the alt text its report suggests, and none if you switch alt text off before you scan.
Legal basis: providing the service you asked for. Retention: the description and the suggested name stay with the result until you delete it; Anthropic's copy at most 30 days in the normal course.
When you create an account
We store your name, your email address, a hash of your password if you set one (never the password itself), the time you verified your email, a new email address you asked to change to until you follow the link we send to it (only then does it replace the old one), and, if you fill them in, a phone number and a profile photo. If you turn on two-factor authentication, we store its secret and your recovery codes, encrypted. If you tick "remember me", or sign in through one of the companies below, we store a token that keeps you signed in. Your history, every compression and page scan with its settings and outcome, and whether an image came through the website, the API, an AI app or a connected WordPress site, is attached to the account.
For our own figures we also note each time an account is created, verified or deleted, starts its Pro trial, connects a WordPress site or was created from the page that connects one, and each time anyone, a guest included, has an image compressed, a page scanned or an AI alt text written. Each note holds the moment, how an image came in (the website, the API, an AI app or a page scan), how a site was connected (from its plugin, from Settings, Sites, or by joining a multisite network), whether a deleted account had been verified and, for an account, its internal number. The note of a compression also holds the plan it ran on, the program that sent it, with its version (Iminify's PHP client or Laravel package when the API request named one, or its WordPress plugin for a connected site's uploads), the internal numbers of the upload group it was one of the files of and of the connected site that sent it, how it ended, how many times it ran and when it first and last finished, and what it measured: the formats and compression level, the file size and dimensions before and after, whether it was resized or fitted under a size limit, and the dimensions and sizes of any extra copies made, such as the widths of a responsive set. A note never holds a name, an email address, an IP address or a file. These notes stay after the account is deleted, and after the image is deleted, so the figures for past weeks don't change. Legal basis: our legitimate interest in knowing how the Service is used.
For security we also keep a record of each signed-in session: its IP address, browser and last activity. You can see them all on the Sessions page in your settings and sign the other ones out, and setting a new password through the reset link we email you signs them all out and drops an email address change still waiting for its link. If the account's address had never been verified, following that link also proves the address is yours, so we verify it and remove the two-factor setup, connected sign-ins, API tokens, connected apps and any address change waiting for its link that anyone set up on the account before. API tokens you create are stored hashed, with the name you gave each one and the time it was last used, which the API Tokens page shows you.
If we ever suspend the account, we record why. If a paid plan is granted to the account, we record the plan and the date it ends, and keep a note of each grant and revocation (which plan was given, until when, and when it changed) under the account's internal number, which stays after the account is deleted. When verifying the address, or the trial button on the website, starts a trial of Pro, we record when it started, so an account gets only one, and when it ends, and we email you two days before the end and on its last day. The trial is also once per person, so we keep a fingerprint of the email address it started on, and of any address the account moves to after it: a SHA-256 hash of the address as its inbox reads it, with any +tag and a Gmail address's dots left out. We keep only the hash, never the address itself, and it stays after the account is deleted, so a later account at the same address, however it's spelt, gets no trial. Each account's own fingerprint is kept with it too, so we can see the accounts one person holds. An address at a disposable email service gets no trial. The trial is once per WordPress site as well: when an account on its trial first opens an upload from a connected site, we keep a fingerprint of the site's host (a SHA-256 hash of its name, without www.) with the account's internal number, and it stays after the account is deleted, so a later trial account sending that site's images gets the free plan's numbers for it. For a site of a WordPress multisite network, we keep the network's host the same way, so the network's sites share one trial. A local copy's host, such as localhost or a .local name, isn't kept.
Signing in with Google, GitHub, Microsoft or Apple. You can sign up or sign in with an account you already have at one of these companies instead of a password. We send you to its page, and if you agree there, it sends us an identifier for your account with it, your name, your email address, whether it has verified that address, and the basic profile details it shares with every sign-in, such as the address of a profile picture. We keep the identifier and the email address it sent, to recognise you the next time, and use the name and email address for your Iminify account; nothing else it sends is kept. It never sends us your password there. If the address belongs to an Iminify account you already have and the company has verified it, you're signed into that account and the two are connected. If that account's address had never been verified, we treat the account as yours: we verify the address and remove the password, two-factor setup, other connected sign-ins, API tokens, connected apps and any address change waiting for its link that anyone set up on it before. You can connect and disconnect these accounts in Settings, Sign-in methods, and choose a password in Settings, Password.
Legal basis: providing the account you asked for, and our legitimate interest in keeping accounts secure and the free trial to one per person and per site. Retention: until you delete the account, except the fingerprints of an address and a site's host that had a trial, which stay after it.
The emails we send you
About the account. We email your account's address the link that verifies it, the password reset links you ask for, the trial's two reminders, a word when your first paid plan starts, and one when a paid plan ends. We also send a notice after each change to the account: a password set or changed, two-factor authentication turned on or off, a sign-in method connected or removed, an app connected, an API token created, a WordPress site connected, a plan given to the account by us, and the account's deletion. When you ask to move to a new address, the current one hears about it, and once the move is made both addresses are told, the old one with the new address in full. When the account is deleted, we send one last email to the address it had: that email, with the address, waits in our mail queue until it goes, and stays in Mailgun's logs for as long as Mailgun keeps them. The notices go only to an address that has been verified, and none of the emails about the account can be turned off while you have one. If an account's address still isn't verified a few days after it signed up, we send it one reminder.
Tips and reports. We also send two kinds of email you can stop: tips, which show you a feature of Iminify and how to use it, and reports about your own images, such as your first results, your month in numbers, a milestone you passed, a daily limit you reached or a site crawl that finished. Each one, and the reminder to verify an address, has an unsubscribe link and the header a mail app needs to unsubscribe in one click, and Settings, Emails turns the tips or the reports off or back on. To choose which to send and when, we read what the account has done: when it signed up and verified its address, its images, the settings they were made with and its page scans and how much they saved, whether it has made an API token, connected an app or connected a WordPress site, which plan it's on and its payments, whether it has run a site crawl, set up a weekly check or a webhook, or shared a report, the last daily limit it reached and when, the days it used Iminify, and when it was last active (an image, a page scan, an AI alt text, or a call from an API token, a connected app or a connected WordPress site). We keep a record of each of these emails we send, and of the reminder and the emails when your first paid plan starts or a paid plan ends: which one, when, whether Mailgun gave up delivering it, and whether you unsubscribed from it or reported it as spam, so that none goes twice and a reminder that keeps failing stops.
No email carries a tracking pixel, so we don't learn whether you opened it, and its links go straight to our site rather than through a service that counts clicks. A link to one of our pages carries tags naming the email it came from, which Google Analytics records with your visit if its cookies are allowed (see "Cookies and similar technologies").
When an email can't be delivered. If Mailgun tells us an email to your address bounced for good, we stop the tips, the reports and the weekly check emails to it until you change the address. If it tells us you reported one as spam, we turn off the tips and the reports, which you can turn back on in Settings, Emails. Either way, Mailgun itself stops every later email to that address, those about the account included, until we ask it to lift the block, which we do when you write to us.
Legal basis: providing the account you asked for, for the emails about it; our legitimate interest in helping you get the most from the Service, for the tips and the reports, which you can stop at any time. Retention: the record of each email, your choices and a bounce mark stay with the account and go when it's deleted.
When you buy Pro or Max
The checkout belongs to Lemon Squeezy, our merchant of record. Lemon Squeezy collects your name, email address, billing address, payment details and any tax information the checkout asks for, and it emails you the receipt. It sends us a customer reference, the subscription's plan, status, interval and renewal date, the card brand and its last four digits, the amount of each order and a link to its receipt, and the amount, tax and date of every payment, renewals included, of any refund and of each failed attempt to charge a renewal. We store those to know which plan your account is on and when it ends, to answer billing questions, to handle refunds and payment disputes, and to keep our accounts.
Legal basis: performing the subscription contract, and our legitimate interest in keeping accounts of what we sold; Lemon Squeezy has its own tax and accounting obligations for the invoices it issues. Retention: with your account, except a record of each payment, refund, failed charge attempt and change to the subscription (the amounts, dates, plan, interval and status, Lemon Squeezy's reference numbers and the account's internal number, without your name, email address or card details), which stays after the account is deleted, as Lemon Squeezy's own invoices do; Lemon Squeezy keeps invoices for as long as tax law requires it to.
When you share a result
Each finished result has a Share action in its menu. Nothing happens until you confirm it. When you do, we give the result a random link of its own, record when you made it public, and from then on anyone who has the link can open a page showing the optimized copy, the original and a comparison of the two where a browser can draw the original (a HEIC or TIFF original is offered as a download only), their sizes, dimensions, format, the settings you chose and the date, and can download both files. They can also compress the original with settings of their own, which stores a copy of it as a result of theirs, under their session or account, exactly as if they had uploaded it; that copy is theirs to keep or delete and doesn't end when you stop sharing. No sign-in is needed for any of this. The link isn't listed anywhere on the site, the page tells search engines not to index it, and its images carry the same instruction, but anyone you or a later visitor pass the link to can open it. Stop sharing from the same menu at any time; the page and its images answer "not found" from then on. Deleting the result ends the share too. If you shared as a guest, you can stop sharing only while your session lasts; after that, email us the link and we'll switch it off.
Legal basis: providing the service you asked for. Retention: until you stop sharing or delete the result.
When you scan a page
We store the address you entered, the addresses of the images found on it with what the scan measured about each (its size, its format and how the page shows it), the images themselves as compressions, and any note about why a scan or an image failed. Our servers fetch the page and its images, so the site you scan sees our address, a browser signature, and the page address as the referring page. It never sees you.
On a paid plan you can crawl a site: give us the address of its sitemap, or a list of pages on one site, and each page is scanned and stored as a page scan, as above. For the crawl itself we store the sitemap's address or the pages you listed, the pages it took and how many more it left out, the settings its images are compressed with, and any note about why it failed. Our servers fetch the sitemap the way they fetch a page, so its site sees our address, not yours. An image the pages share is compressed and stored once, unless two pages reach it at the same moment.
On a paid plan you can have a site you've crawled checked every week. For the check we store the sitemap's address or the pages of the crawl it was made from, the settings, whether its emails are on and when it runs next. Each week our servers fetch the sitemap or the pages again, and every image on them, the way a crawl does, so the site sees our address, not yours, and each week's check is stored as a site crawl, as above, keeping only the images that are new or changed since an earlier check. When a check finds any, or can't scan a single page, we email your account's address: the site, the pages with new or changed images and how many, a link to a zip of those images that works for two weeks for anyone who has it, and a link that stops the check's emails. Following that link, or your mail app's unsubscribe button, turns the emails off and leaves the check running.
On a paid plan you can share a scan's report. Nothing happens until you confirm it. When you do, we give the report a random link of its own and record when you made it public, and from then on anyone who has the link can open a page showing the report: the scanned page's address, the addresses and names of its images, what the scan measured about each, the AI alt text suggested for an image shown without one, and thumbnails of the compressed copies. No sign-in is needed, and nothing on the page can be downloaded or changed. The link isn't listed anywhere on the site, and the page and its thumbnails tell search engines not to index them. It answers "not found" while your account isn't on a paid plan, and for good once you stop sharing, on any plan, or delete the scan. A report's CSV and PDF are made when you download them; we don't keep a copy.
Legal basis: providing the service you asked for. Retention: until you delete the scan, or the crawl a page scan belongs to, or stop the weekly check that made the crawl; a weekly check until you stop it; a report's public link until you stop sharing or delete the scan.
When you use the API
The API does what the website does, with the token you send standing in for your signed-in browser, so everything above applies to it in the same way: an image you send is stored and kept like an upload and gets an AI alt text unless the request switches it off (an upload group gets one only when its request asks for it), a scan is kept like a scan, both are attached to your account, and both count against its daily limits. Each request carries your IP address and your token; the limits on requests per minute are counted per account. If a request names the program that sent it in an Iminify-Client header, and that program is one of Iminify's own (its PHP client or its Laravel package), we keep the name and the version with the image; any other name is ignored.
An upload group is one image sent through the API with the sizes and formats to make from it, the way a website keeps several sizes of each upload. For a group we store the name you gave the image, what you told us about it (its format, file size, width and height), each size you asked for (its name, its width and height, how to crop it, and the file size of the copy you already have) and the formats, then the image itself once you send it and every file made from it. A group's files are kept like an upload's and listed only in the group. If the image never arrives, we delete the group about an hour after it was opened.
An image or an upload group sent through the API can ask to be deleted once you've saved it (ephemeral). It's kept like any upload until you delete it, and if you haven't by about a day after it finished, we delete it ourselves: an image with its original and its optimized copy, a group with its image and every file made from it.
When you give the API the address of an image instead of the file, our servers download it from that address, so the site that hosts it sees our address, and its own address as the referring page, never yours. The file is stored like any upload and named after the address, and the address is kept with it, so the result can say where it came from.
Legal basis: providing the service you asked for. Retention: as for uploads and page scans above, except an image or upload group sent as ephemeral, which goes about a day after it finished at the latest, and an upload group whose image never arrived, which goes about an hour after it was opened; a token until you delete it or the account.
When you connect an AI app
Apps like Claude, ChatGPT and Cursor can use Iminify for you through our MCP server, which is another way into the API. Connecting one starts with the app registering itself with us, which stores the name it gives and the addresses it asks us to send you back to. You then approve it on a page of ours while signed in. The app gets an access token that works for an hour, and a refresh token that it trades for a new pair and that lasts 30 days. We store both, the app they belong to, and when the app last called, which Settings > Connected apps shows you. Disconnecting an app there stops its tokens at once. An app can also be set up with one of your API tokens instead, and then it's covered by the section above.
What the app does with your images is stored and counted exactly as for the API. When it sends one of your files to an upload link, we keep the file until the app compresses it, and delete it about a day after it arrived if it never does. The download links the app gets for your images work without signing in for an hour, so anyone who has one can download that image until it expires.
What the app asks us for, we send to it: your images' names, folders, sizes, settings and links, their AI alt text and suggested file names, the addresses they were fetched from, the addresses of the pages you scan, and your plan and today's usage. Your name and email address aren't among them. The app, and the company that makes it, handle what they receive under their own terms and privacy policy, not ours, and they aren't our processors: you choose to connect them.
Legal basis: providing the service you asked for. Retention: tokens until they expire or you disconnect the app, then removed within a day (an expired one 31 days after it expired); files sent to an upload link about a day.
When you connect a WordPress site
Our WordPress plugin connects a site to your account. When you press Connect in it, the site's server sends us the site's home address and its name, and we keep them for the connection, with a hash of a secret its server keeps. You open our connect page and sign in or sign up there, and the account you're signed in with is noted on the connection; once you approve the site, its server collects a key made for the site with its secret, and we email you that the site was connected. We delete each connection within about two and a half hours of its start, whether or not its key was collected.
For a connected site we store its address, its name, the settings its uploads follow, its key (stored hashed, like an API token) and when it last called us, which Settings, Sites shows you with what the site has saved. Every call the plugin makes carries the site's home address, the plugin's version and the site's IP address. To estimate how long its bulk run takes, it also sends how many images and files it still has to send. If you ask it to write alt text for images already in your library, it sends us a small copy of each one with no alt text. If you choose in the plugin to use the settings another image optimizer kept on the site, it sends us those settings, that optimizer's name and the names of the image sizes the site makes, and we answer with the settings they match without keeping any of it: nothing changes until you confirm them. When you audit one of the site's pages from the plugin, it sends us that page's address, which must be on the site's own host, and our servers scan the page as a page scan does (see "When you scan a page"), counted against your page scans, but compress none of its images and write no alt text; we keep the audit with the site, and only the site's key reads it. The site's images come as upload groups, stored and counted as the API section above describes, and each is deleted about a day after it finished, once the plugin has saved its files. The key works only for that site's uploads, audits and settings, and to read what the plugin shows of your account (your plan, today's counts, and how many images we optimized and bytes we saved, for the site and for your whole account), and only from the site's own address, except to move the connection to the site's new address. Disconnecting the site in Settings, Sites, or from the plugin, stops the key at once; the site stays listed, disconnected, until you remove it.
A WordPress multisite network is connected once, through its main site, and the connect page says it's a network before you approve it. Each of its other sites joins the first time it needs us: the network's server sends us that site's home address and name with the network's key, and the site gets a key of its own, stored and limited as above, starting with the network's settings. We email you when the network is connected, not as each site joins. Disconnecting the network stops its sites' keys too.
What the plugin receives from us is the site's settings, your plan, today's usage and the estimate, the files made from its images, the alt text written for them, and the audits of its pages. It runs on your WordPress site, under your control, so what happens to them there is up to you.
Legal basis: providing the service you asked for. Retention: a connection, used or not, about two and a half hours at most; a connected site, with the audits of its pages, until you remove it or delete the account, its key until you disconnect it.
When you set up webhooks
On a paid plan you can give us the address of your own server in Settings > Webhooks, and we then send it a request each time one of your images or page scans finishes or fails, and when an image's responsive sizes are done. We store the address, a secret that signs those requests (you see it only when it's made), and when the last request went out and what your server answered. The address has to use https and lead to a server on the public internet.
Each request carries what the API answers about that image or scan: its names, folder, sizes, settings and status, its AI alt text and suggested file name, the address it was fetched from or the address of the page you scanned, and download links that work only with your API token. Your name and email address aren't in it. Whoever runs the server you name receives it under their own terms, not ours, and isn't our processor: you choose where it goes.
Legal basis: providing the service you asked for. Retention: the address, the secret and the last request's outcome until you remove them or delete the account. While the account isn't on a paid plan, nothing is sent and they're kept.
When you email us
We keep your address and what you wrote for as long as it takes to deal with it, and for a reasonable time afterwards in case the matter comes back.
Your images
Compression is automated. Software decodes the image, re-encodes it, and measures the result; nobody looks at what an image shows, and no outside service is ever sent an image to compress it. Every encoder we use runs on the machines that also store your files. The one copy that leaves is the small one AI alt text sends to Anthropic, described in "When AI writes your alt text", and only while that setting is on.
Your images are private unless you share one. A guest reaches them through the browser session that uploaded them; an account holder reaches them by signing in. A result you've shared is the exception you chose: anyone with its link can open it until you stop sharing (see "When you share a result"), and so is a page report you've shared, whose thumbnails show the scan's compressed copies (see "When you scan a page"). The other exception is us: the admin tools we use to run the Service show a thumbnail of every stored image and let us open the full-size file, and they let us sign in as an account to reproduce a problem the account holder has reported. We use that access to support you, to investigate abuse or a report about content, and when the law requires it. Files you download pass through Cloudflare on the way to you, over an encrypted connection.
By default the optimized copy has its EXIF metadata removed: camera details, location, dates. If you choose to keep it, the copy keeps it. The original file you uploaded keeps its metadata in storage either way. If the upload carries Content Credentials (a signed record of where the image came from, such as the one Google's Gemini adds), we note that it did and whether the record says generative AI made it, and show you on the result whether the optimized copy still carries them.
When we make a file at Smart or Ultra (an optimized copy, or one of the widths or sizes made from it), we keep a fingerprint of it: a keyed hash of its bytes, which can't be turned back into the picture and isn't linked to you, your account or the image's record. If that exact file is sent to us again at the level it was made at (or at Smart, for a file made at Ultra), without a resize, a size limit it's over or another format (Auto doesn't count as one), we hand it back as it is rather than compress it a second time, which would cost it visible quality. Anyone holding that exact file could learn from the answer that we made it, though not for whom or when. The fingerprint stays after the image and the account are deleted, since nothing in it leads back to either, and it covers the files we made before this policy's date too. Legal basis: our legitimate interest in not degrading the files we've already optimized.
An image may show other people. We don't know who they are, so we can't tell them their picture passed through Iminify. If you process images for a business or another organisation and they contain other people's personal data, that organisation is responsible for the data and we act as its processor; a data processing agreement is available on request.
Retention and deletion. Originals and optimized copies stay in storage until they're deleted. We don't delete them on a timer, except an image or upload group sent through the API that asked us to (ephemeral), which we delete about a day after it finished, and an upload group whose image never arrived, which we delete about an hour after it was opened. Each image, page scan and site crawl in your results has a delete button; using it removes the record, the original and every optimized copy at once (only the note of what the compression measured stays, for our own figures, as described under "When you create an account", and the fingerprint of a file we made at Smart or Ultra, described under "Your images"), deleting a scan removes every image it found, deleting a crawl removes every page it scanned with their images, and stopping a weekly check removes every crawl it made. Deleting your account removes all of them. Deletion is permanent: once a file has left storage, neither you nor we can get it back. If a file reaches the upload box but the page is closed before it's compressed, it never becomes a result, and we delete it about an hour after it arrived.
If you compressed images as a guest, your access to them ends with your browser session, about two hours after your last request or when you clear your cookies. The files stay in storage after that, unreachable by anyone but us, until we remove them. Email us the file name and roughly when you uploaded it and we'll delete it.
Cookies and similar technologies
Cookies are small files a website stores in your browser. These are the ones Iminify uses:
| Cookie | Set by | What it does | Lasts | Needs consent |
|---|---|---|---|---|
iminify_session |
Iminify | Keeps you signed in, and holds a guest's results, your AI alt text settings, whether your uploads start on their own from one page to the next, and the WordPress site you're connecting while you sign in or sign up | 2 hours after your last request | No |
XSRF-TOKEN |
Iminify | Protects forms against cross-site request forgery | 2 hours | No |
remember_web_… |
Iminify | Keeps you signed in when you tick "remember me" or sign in with Google, GitHub, Microsoft or Apple | Until you sign out, at most 400 days | No |
iminify_consent |
Iminify | Remembers the choice you made in the cookie banner | 180 days | No |
theme and flux.appearance (browser storage, not cookies) |
Iminify | Your light or dark mode choice. flux.appearance is kept only while you've picked Light or Dark, and choosing System removes it |
Until you clear it | No |
flux-sidebar-collapsed-desktop (browser storage, not a cookie) |
Iminify | Set on your first visit by the site's menu, to remember whether it's folded away on a wide screen. It holds nothing about you | Until you clear it | No |
iminify:docs-language (browser storage, not a cookie) |
Iminify | The programming language you picked for the code samples in the API docs | Until you clear it | No |
iminify_theme |
Iminify | Tells our server whether you're in light or dark mode, so every page arrives already drawn in it | 1 year | No |
cf_clearance |
Cloudflare | Records that you passed a security check, so you aren't asked again | 30 minutes | No |
ls_aff_ref |
Lemon Squeezy | Set only when you arrive through an affiliate's link: remembers that visit so a purchase is credited to the affiliate | The referral period Lemon Squeezy sets | Not asked |
_gcl_au and other _gcl_ cookies |
Google Ads | Link a click on one of our ads to what you then do on the site | 90 days | Yes |
_ga, _ga_… |
Google Analytics | Tell one visitor from another and count visits | 2 years | Yes |
IDE, test_cookie and others on doubleclick.net and google.com |
Remarketing: showing our ads to people who have visited | Up to 13 months | Yes |
The cookies that need consent come from Google's tags, which Google Tag Manager loads for us: the Google Ads tag measures our advertising, and Google Analytics shows how the site is used. Through them we report the pages you open and these moments:
- you create an account, and whether you used a password or Google, GitHub, Microsoft or Apple, or verify its email address;
- an image is queued for compression, with its file type, the compression level, the output format, whether it's resized, and the language of its AI alt text or that alt text is off;
- you copy an AI alt text, or use the file name it suggests;
- a page scan is queued;
- you download a result, and how many images the download holds, or someone downloads a file from a shared page;
- you make a result public, and each time its link is copied, handed to your device's own share sheet, or sent to a social network or by email from the site, with which one;
- you reach the daily image, scan, site crawl or AI alt text limit, or an upload is too large for your plan;
- you open the checkout for Pro or Max, with the plan, the billing interval and its price;
- you buy Pro or Max, with the order number, the amount before tax, the currency, the plan and the interval.
Each page view also says which plan you're on: guest, free, Pro trial, Pro or Max. None of these reports carries your images, file names, name or email address, or the address of a page you scan.
Google Analytics also notes a few things on its own: when you scroll to near the end of a page, when you start or send a form, when you follow a link to another site, and when you download a page scan report as a CSV or PDF file. For a link or a report it records the full address of what you clicked. On the share buttons of a public result, that address holds the result's public link and the text that goes with it, which names the file. In a page scan report, it's the address of the scanned page or image you open, or of the report file you download.
We don't load Tag Manager at all on the password reset page, because its address carries your reset link.
How consent works. If you're in the European Economic Area, the United Kingdom or Switzerland, Tag Manager and Google's tags don't load until you press Accept in the cookie banner. Reject is right next to it and does what it says; nothing about the site changes either way. Anywhere else they load unless you've said no. You can change your mind at any time from "Cookie settings" in the footer of every page, which turns them off and removes the Google cookies we can reach. If your browser sends the Global Privacy Control signal, we treat it as a no and don't ask.
Affiliate links. Some people recommend Iminify through Lemon Squeezy's affiliate programme, with links that carry ?aff= and a code. When you arrive through one, a Lemon Squeezy script on our pages works out an identifier from your browser's characteristics, sends it to Lemon Squeezy with the address of the page and the page you came from, and sets the ls_aff_ref cookie. If you then buy a paid plan, we pass that reference on to the checkout so the affiliate is credited. The script itself is downloaded from Lemon Squeezy on every page, which hands them your IP address and your browser's identification string the way any download does; on a visit that didn't come through an affiliate link and has no such cookie, it records nothing more. It runs whatever you chose in the cookie banner, but never on the password reset page or the email verification link, because their addresses carry a secret.
Third-party vendors, including Google, use cookies to show ads based on someone's earlier visits to our site. Google explains what it does with the data it collects on sites like ours at How Google uses information from sites or apps that use our services. You can switch off personalised ads from Google in Google's ad settings, and from other vendors on the Network Advertising Initiative's opt-out page.
Who we share data with
We share personal data only with the companies that help us run the Service, listed below, each of which is bound by a data processing agreement, and with authorities when the law requires it. If Iminify is ever sold, its data would go to the new owner under this policy, and we'd tell you first.
| Provider | What it does for us | What it receives | Where |
|---|---|---|---|
| Hetzner Online GmbH | Runs our servers and object storage | Everything the Service stores, including your images | Germany |
| Cloudflare, Inc. | Sits in front of the site: DNS, TLS, protection against attacks, page load measurement | IP address and request data, page timings measured in your browser | United States, with a global network; certified under the EU-U.S. Data Privacy Framework |
| Google LLC | Google Tag Manager, advertising measurement and Google Analytics | Pages opened, the events listed under cookies above, the full addresses of links you follow to other sites and of page scan reports you download, cookie identifiers, IP address, device details; only with your consent where it's required | United States; certified under the EU-U.S. Data Privacy Framework |
| Lemon Squeezy (Sold through Link, LLC) | Sells Pro and Max as merchant of record, takes the payment and runs our affiliate programme | Your name, email address, billing address, payment details and tax information; on every page, the IP address and browser identification string of the affiliate script's download; if you arrive through an affiliate's link, an identifier worked out from your browser, the page address and the page you came from | United States; standard contractual clauses |
| Mailgun Technologies, Inc., a Sinch company | Delivers the emails we send: verification, password resets, notices about your account, trial reminders, weekly check reports, tips and reports on your images; tells us when one bounces for good or is reported as spam | Your email address, the contents of those emails, and which of our emails each one is | Processed in Mailgun's EU region; certified under the EU-U.S. Data Privacy Framework |
| Anthropic, PBC | Writes the AI alt text and suggested file names | A small copy of each image you compress while AI alt text is on, or that a connected WordPress site asks one for (at most 768 pixels on its longest side, without metadata) and the language you picked | United States; standard contractual clauses |
| Flare (Facade BV) | Tells us when something breaks and how fast the site is | Error reports and timing samples from a small share of requests: the address and route, your IP address and browser, the signed-in account's id, name and email address, the form fields you sent and the surrounding lines of our own code, never passwords, sign-in or reset codes, access tokens, cookies or session contents. Our application's log entries, each with the full address of the request that wrote it | Facade BV is a Belgian company, in the European Union |
Each of them publishes its own privacy policy: Hetzner, Cloudflare, Google, Lemon Squeezy, Mailgun, Anthropic and Flare.
When you sign in with Google, GitHub, Microsoft or Apple, that company isn't our processor: it runs the sign-in for you under its own privacy policy and records it there (Google, GitHub, Microsoft, Apple). We send it nothing about you beyond the request to sign you in.
We don't sell personal data, and we don't share it for advertising beyond the Google tags described above.
Where your data is processed
Our servers and our image storage are in the European Union, in Germany. Email is sent through Mailgun's EU region, and error reports go to Flare, run by a Belgian company. Cloudflare handles requests at the network location nearest to you. The small copies AI alt text sends to Anthropic are processed in the United States.
Iminify is operated from outside the European Union, from a country that isn't covered by an EU adequacy decision, so the admin tools are read from there. That access is limited to what running the Service needs, happens over encrypted connections, and is covered by the commitments in this policy; it's necessary to provide the service you asked for (Article 49(1)(b) GDPR).
The providers based in the United States receive data under the EU-U.S. Data Privacy Framework where they're certified, and under the European Commission's standard contractual clauses otherwise, as listed in the table above.
How long we keep things
| Data | Kept for |
|---|---|
| Uploaded images, optimized copies, page scans, site crawls, weekly checks | Until you delete them, or until the account is deleted; an image sent through the API as ephemeral, about a day after it finished at the latest |
| Upload groups: the image's name, what you said about it, the sizes and formats you asked for with your copies' file sizes, the image and every file made from it | Until you delete the group, or until the account is deleted; a group sent as ephemeral, about a day after its last file finished at the latest; a group whose image never arrived, about an hour after it was opened |
| Images added to the upload box but never compressed | About an hour |
| AI alt text and suggested file names | With the result they describe, until it's deleted |
| The small copy made for AI alt text | On our server until the answer arrives, at most a day; at Anthropic at most 30 days, longer only if its safety systems flag it |
| An AI alt text a connected WordPress site asked for on an image already in its library, with its language and the fingerprint of the copy | A day from when it was asked for |
| Guest results | Files until deleted; your access ends about 2 hours after your last request |
| A fingerprint (a keyed hash, never the picture) of each file we made at Smart or Ultra | Kept after the image and the account are deleted, so the same file sent again isn't compressed twice |
| Public links to shared results | Until you stop sharing or delete the result |
| Public links to shared page reports | Until you stop sharing or delete the scan; paused while the account isn't on a paid plan |
| Account details and history | Until you delete the account |
| A fingerprint (a hash, never the address) of the email address a Pro trial started on, and of any address the account moved to after it | Kept after the account is deleted, so the trial isn't taken again |
| Your email choices, a bounce mark, the last daily limit reached, and a record of each tip, report and reminder we emailed you (which one, when, whether it could be delivered, and whether you unsubscribed or reported it as spam) | Until the account is deleted |
| The last email to a deleted account's address | In our mail queue until it's sent; in Mailgun's logs for as long as Mailgun keeps them |
| Connected sign-in accounts (the identifier and email address a company sent) | Until you disconnect them, or until the account is deleted |
| API tokens | Until you delete them, or until the account is deleted |
| A WordPress site waiting to be connected (its address and name, a hash of its secret and the account that opened the connect page) | About two and a half hours from its start at most, whether or not its key was collected |
| A connected WordPress site (its address, name, settings, key and when it last called) | Until you remove it, or until the account is deleted; its key until you disconnect it |
| The audits of a connected WordPress site's pages (each page's address and what the scan measured about its images) | Until you remove the site, or until the account is deleted |
| A fingerprint (a hash) of a WordPress site's host where a Pro trial sent its images | Kept after the account is deleted, so the site's trial isn't taken again |
| A webhook address, its signing secret and its last request's outcome | Until you remove them, or until the account is deleted |
| Connected AI apps' tokens | An hour (access) and 30 days (refresh) of use; removed within a day of disconnecting the app, 31 days after expiring, or with the account |
| Files an AI app sent to an upload link | Until the app compresses them, or about a day |
| An AI app's registration (the name and callback addresses it gave, the same for everyone who connects that app) | Until a day after the last token issued to it is removed |
| Signed-in sessions | 2 hours after the last request; expired ones are cleared |
| Password reset links | 1 hour |
| Daily usage counters | 24 hours |
| Application logs | 14 days |
| Records of failed background jobs, which carry a file name | 7 days |
| Error reports, timing samples and log entries at Flare | 30 days |
| Billing records from Lemon Squeezy | With the account; a record of each payment, refund, failed charge attempt and subscription change, and of a paid plan granted by hand, without your name or email address, stays after it; Lemon Squeezy keeps its invoices as long as tax law requires |
| Notes of sign-ups, verifications, deletions, trials, connected sites, compressions, page scans and AI alt texts (the moment, the way in and the account's internal number; for a compression, also its plan, the program of ours that sent it, its upload group's and its connected site's numbers, how it ended, how many times it ran and what it measured) | Kept after the account or the image is deleted, for our own figures |
| Google Analytics data | As set in our Google Analytics account, at most 14 months |
Security
Every connection to the site is encrypted, and Cloudflare filters attacks before they reach our servers. Passwords are hashed with bcrypt and never stored in the clear. Two-factor authentication is available in your settings, and its secrets are encrypted at rest. Access to the servers, the storage and the admin tools is limited to those who run the Service. Our software dependencies are checked against published security advisories, and the Service is monitored around the clock.
If a breach ever puts your data at risk, we'll tell the supervisory authority within 72 hours where the law requires it, and we'll tell you without undue delay when the risk to you is high.
Your rights
You can do most things yourself: change your name, email address, phone number and photo in Settings; connect or disconnect the accounts you sign in with; delete any image, scan or site crawl from your results; stop a weekly check, or only its emails; turn the tips and the reports off in Settings, Emails or from any of them; delete your account from Settings, Profile, Delete Account; and change your cookie choice from the footer.
You also have the right to ask us to:
- confirm what personal data we hold about you and give you a copy of it;
- correct anything that's wrong;
- delete your data;
- restrict or object to how we use it;
- give you your data in a portable format;
- withdraw a consent you gave, without affecting what was done before.
Email [email protected] from the address on your account, or, as a guest, tell us what you uploaded and when. We answer within a month; if a request is unusually complex we may take up to two more months and will say so. We'll ask you to prove who you are only where we need to.
Nothing here is required by law. Without an image there's nothing to compress, without an email address there's no account, and without a payment there's no paid plan; everything else, the phone number, the photo, the advertising cookies, is optional. We make no decisions about you by automated means that have legal or similarly significant effects.
You can also complain to us directly at [email protected]. We acknowledge a complaint within 30 days and answer it without undue delay.
If you think we've handled your data unlawfully, you can complain to a supervisory authority: in the EU, the authority of the country where you live or work; in the United Kingdom, the Information Commissioner's Office. We'd appreciate the chance to sort it out first.
If you're in California or another US state with a privacy law, we don't sell personal information and we don't share it for cross-context behavioural advertising except through Google's tags, which you can switch off from the footer or with the Global Privacy Control signal. The rights above are available to you in the same way.
Children
Iminify isn't directed at children. You have to be at least 16 to use it, we don't knowingly collect data from anyone younger, and if we learn that we have, we delete it and close the account. If you think a child has given us their data, email us.
Changes to this policy
The date at the top is the date this version took effect. For a change that affects what we collect or how we use it, we'll email account holders before it takes effect and post a notice on the site. Earlier versions are available on request.
Contact
Email: [email protected]
Website: https://www.iminify.com