Legal documents
Privacy Policy
Revised 2026-09-04
- •We keep your email, billing reference and the videos you process — nothing beyond what the service requires.
- •No advertising trackers: audience measurement is first-party, served from our own domain, and stays switched off until you accept it in the banner. BYOK API keys are obfuscated inside your own browser and never reach our servers.
- •Your content is never used to train AI models. Free-plan clips are erased after 7 days.
1. Who is accountable
The controller of your data is TONVI TECH SL (CIF B-19780394), Calle Puerta del Mar 18, 5th floor, 29005 Málaga, Spain — the company behind Clipping AI (clip.zes.codes). On any data matter, contact [email protected]. This policy also exists in Spanish; for people living in Spain, the Spanish text prevails if the versions differ.
2. What we handle, why, and on what legal ground
Your account
We hold your email address (plus your Google account identifier if you log in through Google), the date you signed up and the date of your last login. Sign-in uses passwordless magic links, so no password is ever stored. Ground: performance of the contract (Art. 6.1.b GDPR).
Billing
All payments are handled by Stripe: we keep your Stripe customer reference, your plan and your subscription status, never your card number. Invoicing records are retained because tax law demands it. Ground: contract (Art. 6.1.b) and legal obligation (Art. 6.1.c).
The videos you process
The videos you upload or ask us to fetch, their transcripts and the clips made from them are handled to carry out the precise job you requested. Ground: contract (Art. 6.1.b). When a video contains other people's personal data (their image or voice), you are the controller of that data and we act as your processor on your instructions — see Section 6 of the Terms of Service.
Your rights declaration
Every time you submit a job you confirm you hold the rights to the content. We retain that declaration with its date, your browser identifier and your IP address truncated to its network (the final part is discarded, so it points to a network rather than a device), as proof of due diligence if a copyright or image-rights dispute arises. It is stored alongside its project and deleted with your account. Ground: legitimate interest in establishing and defending legal claims (Art. 6.1.f).
Service emails, and the one that is not
Magic links, receipts, renewal reminders, clip-ready notices and warnings ahead of deletion all arrive by email. They form part of the service, not marketing. Ground: contract (Art. 6.1.b). One message is different: if you are on the free plan and exhaust your minutes, we send a single email suggesting an upgrade. That is a commercial communication (Art. 21.2 of Spanish Law 34/2002, which allows it for our own similar products to our own customers); it includes an unsubscribe link and a one-click unsubscribe header in every send, and opting out stops it permanently without affecting service messages. We run no newsletter and never sell or rent your address.
Product analytics
We see how the app is used through OpenPanel, running on an instance we operate ourselves; the measurement script comes from this domain rather than a third party. Nothing loads until you accept the "audience measurement" category in the cookie banner, and refusing takes one click in that same banner — the "cookies" link in the footer lets you change your mind at any point. These events carry the account's internal identifier (a random uuid) so we can distinguish first-time from repeat use; your email address is not among them. No advertising pixels, no cross-site tracking, nothing passed on to anyone. Ground: your consent (Art. 6.1.a GDPR and Art. 22.2 of Law 34/2002); withdraw it whenever you like in the banner or by emailing [email protected].
Security and anti-abuse
Server logs, rate limits based on IP and fraud signals are kept to shield the service and its users. Ground: legitimate interest (Art. 6.1.f). Application logs remain in the container's rotating log for at most 7 days, are not gathered anywhere else and contain no email address.
We do not handle special categories of data (Art. 9 GDPR) as part of the service, we take no automated decisions with legal effects on you (Art. 22 GDPR), and we do not use your content or your data to train AI models — ours or anyone else's — nor sell it.
3. Cookies and local storage
We place no third-party cookies and pull in no third-party script when a page loads. What the site stores falls into two groups.
Strictly necessary — always on, no consent needed. It is all first-party and lives in your browser's local storage rather than a cookie: your session token (keeps you signed in), a short-lived media token (lets your browser fetch your own clips and thumbnails, which are not public), your interface preferences (last-used editor settings, tab, language), the first-visit referrer recorded once at sign-up, and — if you use bring-your-own-key mode — your AI provider API keys. Those keys are obfuscated, not encrypted: the value is scrambled so it cannot be read at a glance in developer tools, which is not cryptographic protection, and anyone with access to your browser profile could recover them. They are sent only to carry out your own jobs and are never stored on our servers. Your consent record itself is also kept here.
Audience measurement — always on, first-party only. OpenPanel (see section 2) writes an anonymous visitor identifier. The script is served from this domain; the instance is ours, the data is never shared and never used across sites, which under the AEPD/CNIL criteria exempts it from prior consent. The consent banner appears only to visitors in the EEA, the UK and Switzerland; identifiers last no longer than 13 months and raw data no longer than 25 months. "Reject all" and "Accept all" sit side by side in the banner, and the "cookies" link in the footer reopens it so withdrawing consent costs no more effort than giving it.
Marketing. We use no advertising or remarketing trackers. The category is present in the banner so that, should we ever add one, it begins switched off.
4. Who receives data (processors) and where
Below are all the providers we rely on, what each one is given and its location, including the transfer safeguard for each. We can send the full list on request.
- Google (Gemini API) — AI analysis to select moments, layouts, titles and thumbnails. It receives your video's transcript and sampled frames (which may show faces), the reference photo you pick for a thumbnail, and — only when the "screencast" layout check runs — the source video file itself, which we delete from Google's file store as soon as the answer comes back rather than letting it expire there. Google's paid API terms prohibit using customer data to train models. USA.
- ElevenLabs — AI dubbing and AI voices, only when you ask for it. Receives the clip's audio, which means the voice of whoever is speaking in it. USA.
- Pollinations — the image and video models behind AI Shorts (Flux for actor images, audio-driven video models for the talking avatar). Receives the actor photo you upload or generate — which may be a photograph of a real person — and the audio to lip-sync it to. USA.
- Stripe — payments and invoicing. USA/EU.
- Cloudflare R2 and Amazon Web Services S3 — storage of your clips and project files. EU/USA regions.
- Cloudflare — DNS and CDN in front of the website, so it sees the IP address and request metadata of every visitor. USA/global.
- Hetzner and Contabo — the servers the service runs on. Germany.
- Download proxies (DataImpulse and static ISP proxies) — when a video platform refuses our servers directly, the download is routed through a commercial proxy. It sees the URL of the video and carries its bytes; it receives no account data of yours.
- Upload-Post — publishing to social accounts you connect, only on your instruction, and the custodian of the platform tokens those connections produce. Operated by TONVI TECH SL under this same policy.
- Aikount — our own invoicing system, which issues the Spanish legal invoice for your payment and therefore receives your billing name, address and email from Stripe. Operated by TONVI TECH SL, servers in Germany.
- OpenPanel — first-party audience measurement, exempt from prior consent (section 3). The instance is ours and runs on our own servers in Germany; the script is served from this domain.
- Namecheap Private Email — delivery of service emails. USA.
- Google (sign in with Google) — only if you choose it, and only the "openid email profile" scope, so we learn your address and Google account id and nothing else. USA.
There are two things we deliberately do not send anywhere: internal operational alerts refer to you by the first characters of your account identifier, never by email address or video title; and the face detection that determines the crop runs on our own servers with no biometric template produced.
Whenever a provider handles data outside the European Economic Area, the transfer rests on the EU–US Data Privacy Framework where the provider is certified and, in all cases, on the European Commission's Standard Contractual Clauses (Art. 46.2.c GDPR) as a fallback, plus the provider's technical safeguards. We opt for EU regions wherever the provider provides them.
5. How long we keep things
- Free-plan clips: 7 days, then permanently deleted (we warn you by email first).
- Paid-plan projects and clips: while your subscription is active, plus 7 days.
- Account data: while your account exists; after deletion, only what is needed to meet legal obligations or resolve disputes.
- Invoicing data: 6 years (Spanish commercial law).
- Rights declarations: up to 5 years with the project they belong to, matching the statute of limitations for civil claims, and deleted with your account if you delete it sooner.
- Working files on the processing server (the source download, the working copies): 1 hour after the job finishes. Uploads waiting for a job: 6 hours.
- Generated thumbnails and the video frames sampled to make them: 24 hours after you last downloaded one.
- Photos you upload to drive an AI actor: deleted with the job that used them, within the hour.
- Application logs: 7 days in the container's rotating log.
- Encrypted database backups: 30 days, after which a deletion you made is also gone from every copy.
- Analytics identifiers: 13 months; raw analytics data: 25 months.
The complete version of this table, item by item, is available on request.
6. Your rights
You may request access, rectification, erasure, restriction and portability of your data, and object to processing carried out on legitimate interests — write to [email protected] or [email protected] from the address on your account; we reply within one month. Two of these you can handle yourself, immediately and without asking anyone: Account → Download my data hands you a JSON file containing everything our database holds about you plus a download link for each clip in your library (Art. 20), and Account → Delete account erases it (Art. 17). You need no permission from us to close your account: Account → Delete account in the dashboard does it yourself, straight away and for good, taking your projects, clips, transcripts, API keys and social connections with it and cancelling any active subscription. Only two items survive, both covered in section 5: your invoices, which tax law obliges us to retain, and a record that the deletion took place, keyed to a one-way hash of your email address rather than the address itself. If you think we are mishandling your data, you may complain to the Spanish supervisory authority (AEPD, aepd.es) or to the regulator in your own EU country.
7. California residents (CCPA/CPRA)
If you live in California, this section is your notice at collection. Over the last twelve months we have collected these categories of personal information: identifiers (the email address you sign in with); commercial information (your plan, your minute balance and the Stripe reference for your payments, never your card number); internet activity (first-party analytics about how the app is used); and audio and visual information (the videos you upload or link, their audio, the clips we produce and their transcripts). We collect them to run the service you asked for, to bill it, to keep it secure and to meet our legal obligations, as detailed in section 2. The sources are you and, where you paste a link, the platform you took the video from.
We do not sell your personal information, and we do not share it for cross-context behavioural advertising. We have not done so in the preceding twelve months, and we do not do it with the data of anyone under 16. We disclose data to service providers only, for the purposes and to the recipients listed in section 4, under contracts that stop them using it for anything else.
Our reframing detects faces and subjects inside the picture so it knows where to crop. That happens on our own servers, is never used to recognise or identify anyone, and produces no faceprint or other identifier that could match a person across videos. We do not use sensitive personal information to infer characteristics about you, so the right to limit its use does not arise.
You have the right to know what we collect and to receive a copy of it, to have it corrected, to have it deleted, and not to be discriminated against for exercising any of these rights: we do not offer a worse service, a higher price or fewer features to anyone who does. Exercise them by emailing [email protected] from your account address, or delete everything yourself with Account → Delete account. We verify a request by the control you have over the account email, which is the same credential you sign in with, and we answer within 45 days. An authorised agent may act for you if they provide your written permission.
8. Age
The service is not aimed at children. You must be at least 16, or the age of digital consent in your country if higher, to create an account, and you are asked to confirm that when you sign up. If you believe a minor has created an account, write to [email protected] and we will delete it.
8 bis. AI-generated content
Dubbing, AI Shorts and AI thumbnails create synthetic content. We label
what we generate in a machine-readable form — a metadata tag on video files and
an XMP DigitalSourceType: trainedAlgorithmicMedia marking on
generated images — as required by Art. 50(2) of Regulation (EU) 2024/1689 (the
AI Act). That marking does not always survive a platform re-encoding your
upload, so when you publish synthetic content you are the one who must switch on
the platform's own AI-content label (Art. 50(4)); the app reminds you at the
moment you post. Cloning a voice or animating a photograph of a real person
without their agreement is prohibited by the
Terms of Service, and both features ask you to confirm
you have that agreement before they run.
9. Changes
Should we alter this policy in any significant way, we will let you know by email or in-app before the change takes effect, and the date shown here always reflects the version currently in force.