Data protection · GDPR

Privacy Policy

Last updated 11 October 2026

This policy explains how Martin Schenk S.L. ("we") handles personal data in connection with the MachineWitness archive and this website. We designed the project to touch as little personal data as possible; most of what we record is technical policy files that contain none.

1. Visitors to this website

This is a static website. We use no analytics, no tracking pixels and no advertising, and we build no profiles of visitors. We set one cookie, and only if you answer the language bar or let it close: mw_lang records which language you read the site in, so that the bar does not ask again. It stays in your browser, is never transmitted to us, and is not readable across sites. Under Article 5(3) of the ePrivacy Directive a language preference the visitor has expressed needs no consent banner, which is why you do not see one. To deliver and protect the site, our hosting/CDN provider processes technical connection data (such as your IP address and browser type) for the short time needed to serve the page and defend against attacks. The legal basis is our legitimate interest in a secure, functioning site (Art. 6(1)(f) GDPR). If you email us, we process your message to reply.

If you suggest a domain

The suggestion form on the coverage page asks for a domain, your e-mail address, and your reason. We use them to decide on the suggestion and to reply to you, and for nothing else: no newsletter, no profile, no disclosure to third parties. The legal basis is our legitimate interest in taking suggestions about what to observe, and yours in receiving an answer (Art. 6(1)(f) GDPR). Suggestions are deleted once they have been decided and answered, at the latest after twelve months. The coverage check itself is different and sends no personal data at all: the domain you type is hashed in your browser, and only the hash is transmitted.

If you order an evidence extract

The order form on the evidence extract page asks for a domain, the days that matter to you, your reply address and your billing details, and lets you add a note. We use them to answer the order, to establish its scope, and to invoice it. We do not use them for anything else, and we disclose them to no one. The legal basis is Art. 6(1)(b) GDPR, because the processing is necessary for steps taken at your request before a contract, and Art. 6(1)(c) GDPR for the invoicing records Spanish commercial and tax law requires us to keep. Those records are retained for the statutory period, currently six years; everything else about an order is deleted once the matter is closed, at the latest after twelve months. You receive a written confirmation of what we recorded, so that a mistake in it costs a reply rather than an invoice.

The form is protected by Cloudflare Turnstile, which checks that a person and not a robot is submitting it. Turnstile processes technical connection data for that check; it sets no advertising cookie and builds no profile across sites. The legal basis is our legitimate interest in not having the order queue filled by machines (Art. 6(1)(f) GDPR).

If you buy through the checkout

Daily observation, the point-in-time record and the evidence extract can be paid for online. We ask for your e-mail address and the web addresses you want to order; we also keep the language, your time zone and, if you use one, an invitation code. Payment is handled by our payment provider, Stripe, which receives your payment details, your name, your billing address and, if you give one, your VAT number. We never see your card details. Subscriptions are managed in the payment provider's customer portal, which you reach through a signed link in our e-mails.

We use these data to take payment, to deliver the order, to send you the confirmation, the delivery, the notice before an annual subscription renews and the confirmation of a cancellation, and to issue the invoice. The legal basis is Art. 6(1)(b) GDPR and, for invoicing, Art. 6(1)(c) GDPR. To curb abuse we count submissions per IP address, but we keep only a cryptographic digest of the IP and delete it after 48 hours (Art. 6(1)(f) GDPR).

We keep the purchase record for 90 days and the order for a point-in-time record for 400 days, so that we can deliver it again if you need it. Invoices and billing data are kept for the statutory period, currently six years. The recordings of a point-in-time record become part of the archive (section 2).

If you enter yourself in the AI Yes/No list

In the list you state, for your website or your social media accounts, whether your content may be used to train AI: yes or no. We keep your e-mail address, your country, the language and, for each entry, the address or user name with your choice, the dates. We do not keep your name or your IP address. For each submission we also count, in rough classes only, the kind of device (phone, tablet, computer), the operating system, the browser family and the country the connection comes from, so that we can see where the form works and where it does not. The full browser identifier is not kept; the count names no one and stays as an anonymous figure when you delete your entries. The legal basis is your consent (Art. 6(1)(a) GDPR), which you can withdraw at any time without any disadvantage. You must be at least 16 years old.

Only what you confirm through the link in our e-mail counts; anything unconfirmed is deleted after 7 days. Every e-mail contains a link with which you can change or delete your entries without logging in; when you delete everything, we also delete your address. If you enter a website and it is verified as yours, we observe it daily as part of the archive (section 2); deleting the entry ends that addition, but what has already been observed stays in the archive. Social media accounts are shown as "not verified" and are never published. We publish only figures, never entries, and no figure per country or network below 10. We pass the data to no one, not to AI companies either. A transfer to a register created under EU law would take place only with a new consent that we would ask you for separately.

Besides the e-mails about your entry (the confirmation link, one single reminder if you have not confirmed after about a day, the confirmed entry, check results, the deletion), we send you once, about a day after your confirmation, an e-mail with where the list stands and your personal link for passing it on. You can unsubscribe from it with one click; the e-mails about your entry still come. Your personal link carries a random code; when someone confirms an entry through it, we count that so you can see the number. Neither of you is named to the other. If you invite people by e-mail from the confirmation page, the invitation opens in your own mail program: we neither see nor store those addresses and do not write to them.

Each confirmed entry also goes into the archive's daily seal, but only as a fingerprint (a salted hash) that names no one. You find the key to it (the salt) on your management page once both witnesses have sealed; with it you can prove your entry later. If you delete the entry, we also delete the key; the fingerprint stays in the seal and from then on can no longer be linked to anyone. The seal is published (/log).

2. The archive: data from third-party domains

MachineWitness visits public European domains and stores exactly what they serve at a small, fixed set of addresses: the machine-readable files robots.txt, ai.txt, text-and-data-mining reservation files and llms.txt, and the homepage. Of the homepage it keeps the response headers and the HTML document: in the curated core every day in full; in the broad ring in full once a week and, since 12 October 2026, every day up to the end of its HTML head. For every response it also keeps the TLS certificate chain as the server delivered it (since 15 September 2026; before that the fingerprint and five parsed fields of the server certificate). The files are technical and policy declarations and mostly contain no personal data. A homepage may well contain some, as a homepage usually does: names, contact details, photographs of people. A file can contain some incidentally, for example a contact name, and so can a domain that is itself a person's name.

Purpose

To keep a neutral, tamper-evident record of what each domain served to machines on a given day, so that it can be independently verified years later. We do not evaluate, rank, or profile anyone. A witness, not a detective.

Legal basis

Our legitimate interest, and the public interest, in maintaining a verifiable evidentiary archive of the machine-readable web: Art. 6(1)(f) GDPR. Where archived material is later the subject of an erasure request, the retention of the cryptographic fingerprint rests on Art. 17(3)(e) GDPR, because without it the proof chain of every other observation sealed on the same day would collapse. We have documented the required balancing test and our erasure procedure, and will provide them on request.

Data minimisation and crawler conduct

We deliberately fetch only a handful of tiny, machine-readable files per domain per day. We do not crawl page content, we do not follow links into applications, we do not log in, and we apply hard size and rate limits. Our crawler identifies itself with a contact address and respects each site's robots.txt. See Crawler & contact.

Recipients and transfers

Only cryptographic fingerprints (hashes) of each sealed day leave our systems, to be time-stamped by external services (for example OpenTimestamps and an RFC 3161 time-stamp authority). These fingerprints reveal nothing about their contents. The archived files themselves are not shared or sold.

For the website, the forms, e-mail and payments we work with providers that process data on our behalf: hosting and content delivery, e-mail delivery (from Frankfurt), payments and invoicing. Where one of them processes data outside the European Economic Area, the transfer rests on the European Commission's adequacy decision for the United States (EU-U.S. Data Privacy Framework) or on standard contractual clauses.

Retention

The archive is intended to be permanent: its value lies precisely in being able to show what was served in the past. This is subject to your rights below.

Pages recorded on request

For a point-in-time record, the customer names specific public addresses, subpages included, and we record their content and an image of how they looked. Those pages may contain personal data of third parties, for example names or photographs in a legal notice or on a team page. The legal basis is our legitimate interest, and the customer's, in preserving evidence (Art. 6(1)(f) GDPR); the balancing test is documented. We record only public pages, without logging in, and only the addresses ordered. The customer receives the recordings. Otherwise the same rights and the same "tombstone" apply as for the archive (section 3).

3. Your rights

You have the rights of access, rectification, erasure, restriction, objection, and portability under the GDPR, exercised by writing to contact@machinewitness.eu.

Erasure and the "tombstone". On a legitimate request, we erase the affected content: the stored file is deleted and replaced by a marker (a "tombstone"). The cryptographic hash and proof structure are kept, so the integrity of the surrounding record still holds: we can show that something was removed, without keeping what it was. We can also exclude your domain from all future crawling on request; for that we ask you to show control of the domain (a DNS TXT record or a file at a path we name), so that nobody can have a domain excluded that is not theirs.

You may lodge a complaint with a supervisory authority, in Spain the Agencia Española de Protección de Datos (AEPD), or the authority in your country of residence.

4. Language of this policy