Privacy Notice

TVN Global Limited (trading as LUVI Creator), company number 15019980, registered office 82a James Carter Road, Mildenhall, Suffolk, IP28 7DE, United Kingdom, is the controller of the personal data described in this notice. ICO registration number: ZC248767.

Questions and requests: privacy@luvicreator.com. You can also write to the address above, marked "Privacy".

This notice covers the website www.luvicreator.com, the LUVI web application, the LUVI connector for AI assistants (the "MCP connector"), LuviBot, our emails and our support channels (together, the "Services"). It applies under the UK General Data Protection Regulation and the Data Protection Act 2018 and, where you are in the European Economic Area, the EU GDPR.

Account data. Name, email address, password (stored only as a hash), profile picture and language preference; if you sign in with Google, the profile details Google shares with us (name, email, avatar, Google account identifier).

Content you submit ("Input"). Prompts, uploaded images, video, audio and other files, reference material, model settings, messages and files you send to LuviBot, and the rules and settings you save for LuviBot (for example preferred resolution, models you do not want suggested, and your automatic approval and upper limits). Input may contain personal data about you or others, including faces and voices. You are responsible for having the right to submit it.

Generated content ("Output"). The images, video, audio, 3D assets and text generated for you, stored in your library until you delete them.

Usage and technical data. Which models and features you use, credit spend, generation records (job identifier, model, timestamps, status, cost), storage use, folder and project structure, team membership, device and browser type, IP address, approximate location derived from IP, referring URL, and diagnostic logs including error reports.

Notification data. The list of notifications we send you (Account → Notifications): each notification's type, time, whether you have read it, whether it was sent to your device, the id of the record it is about (conversation, invitation, workflow, generation) and the name of the person, team, project, workflow, model or plan it is about; and your notification settings. If you allow notifications in a browser, also: that browser's notification address and encryption keys, the browser type and the site language; and, while LUVI can notify at least one of your browsers, a short-lived record of whether your open LUVI tabs are visible and which conversation or workflow each one has open. A notification's text can include the first 140 characters of a message or prompt; that text is sent to your device but is not kept in the list.

Payment data. Purchase history, credit balance, invoices and plan status. Card details are collected and stored by Stripe, not by us; we receive only a payment token, the last four digits, card brand and billing country.

Verification data. Only if you choose to verify a phone number or a card to earn bonus credits: for a phone, the last two digits of the number; for a card, its brand, last four digits, issuing country and type (credit, debit or prepaid). For both we also keep a one-way fingerprint that cannot be turned back into the number or the card. We do not store the phone number itself; Prelude, the service that texts you the code, does (see sections 4 and 6). To give the welcome credits once per person, we also keep a one-way fingerprint of the IP address an account first signs in from; for welcome credits checked before we stopped using a browser identifier in October 2026, we also keep a one-way fingerprint of that identifier.

Communications. Support emails and messages, feedback and survey answers.

Analytics data. Only if you consent in the cookie banner: page views, clicks and product events sent to PostHog (see the Cookie Policy), including a few events our servers record, such as a completed payment. If you are signed in, your banner choice is also kept with your account so that these server events follow it. Session recording is switched off.

We do not knowingly collect data from anyone under 18 and do not intentionally collect special category data. If your Input contains special category data (for example health information), we process it only to provide the Services you requested. Faces and voices in your Input are used only to generate the Output you asked for: we do not use them to identify, verify or authenticate anyone, and we do not store any faceprint or voiceprint derived from them.

We do not use your Input or Output to train AI models, and we do not sell personal data. The research above covers text only: the files you upload, the faces and voices in them and your Output are not part of it. Once you delete a LuviBot conversation, a generation or your account, we stop using that data for research. You can object to this research at any time (section 7). It covers text you have sent us since 24 September 2026, when the current version of LuviBot launched. We do not use notification data for research, marketing or model training, and we never send advertising or campaigns as notifications. When an email address is invited to a team, we check whether it belongs to a LUVI account so that the invitation can also appear there as a notification; the person who sent the invitation does not see the result of this check.

We use these processors and, where stated, independent controllers. Each is bound by a contract that meets UK GDPR Article 28.

Model inference — Atlas Cloud AI Inc. (New York, United States). Your prompts and uploaded media are sent to Atlas, which routes them to the third-party model you selected (providers currently include OpenAI, Google, ByteDance/BytePlus, Alibaba Cloud, xAI, MiniMax, Kuaishou, Black Forest Labs, ElevenLabs, Suno and others; the model page names the provider). Atlas states that it retains submitted content for up to 7 days for service delivery and abuse prevention and does not use it to train models. Each underlying provider applies its own retention and usage terms to content routed to it.

LuviBot — DeepSeek (Hangzhou DeepSeek Artificial Intelligence Co., Ltd.) (Hangzhou, China). Your LuviBot messages, the files you attach to them and the relevant parts of your account (for example your projects, library items you refer to and your credit balance) are sent to DeepSeek, which runs the language model behind LuviBot (currently DeepSeek V4.1 Flash) and processes them in China on our behalf, under the terms of its developer platform. If you do not want your content processed this way, do not use LuviBot; the rest of the Services does not depend on it.

Message and image reading, policy check — TypeSafe AI, Inc. (San Francisco, United States), reached through Vercel AI Gateway. An automated classifier reads the text of each message you send to LuviBot (together with your previous message when the latest one is short) to tell what you are asking for — for example whether it is a question about a price, or which help article or workflow template fits — and checks it against our Terms. Before LuviBot prepares a paid generation, it also checks the prompt and a short description of the attached references. The classifier also reads images: those you attach to a LuviBot message, and each image you upload to your library or generate there. It notes what an image shows — for example a product or a person, a photo or an illustration, whether there is text or a plain background, and how many people — so that LuviBot can put each file where it belongs and your library can sort images into categories. Images and message text are sent with the zero data retention setting of Vercel AI Gateway: TypeSafe does not train on them and keeps them only as long as needed to return the result. When the classifier is not sure what an image attached to LuviBot shows, the language model that runs LuviBot (DeepSeek, see above), which already receives that image, reads it instead; for an image in your library the category is simply left empty — library images are not sent to DeepSeek for this. The categories read from your images are stored with your account and deleted when you delete the image or your account. If the classifier cannot answer, the language model that runs LuviBot (see above) checks the prompt instead.

Other AI assistance — Anthropic (Claude models; prompt rewriting, automatic tags on your uploads, captions in workflows and help with the research described in section 3). Only the content needed for the feature you invoked, or for that research, is sent.

Hosting and database — Vercel (application hosting, Frankfurt region) and Supabase (Postgres database and authentication, Frankfurt, Germany).

Content checks — Amazon Web Services (Amazon Rekognition, United States). Each uploaded image, each image used in a generation request and each generated image is sent to Amazon Rekognition to detect nudity and sexual content and to estimate the apparent age of any face it shows. Rekognition is not used to identify who a person is. We have opted out of Amazon storing or using this content to improve its AI services.

Media storage and delivery — Amazon Web Services (S3 storage and CloudFront delivery, United States). Your uploads and Output are stored here, encrypted at rest, and served only to your browser through short-lived signed cookies.

Payments — Stripe (independent controller for payment processing and fraud prevention; see Stripe's privacy policy). Stripe also handles card verification for the card bonus, without charging the card.

Phone verification — Prelude (Ding Negoce SAS, Paris, France; SMS verification). Only if you choose to verify a phone number: the number, your IP address, browser details, the site language and an internal reference for the request are sent to Prelude, which texts you the code through telecommunications operators, checks it and screens the request for SMS fraud. Prelude keeps the number, the IP address and the browser details together with a record of each verification: when it was made, the mobile carrier, whether the text was delivered and whether the code was correct. Prelude also uses this data on its own account to fight fraud and to improve its services, and for that use it is an independent controller (see Prelude's privacy policy).

Bot protection — Cloudflare (Cloudflare, Inc., San Francisco, United States). When you sign up or sign in with email and password, reset your password, or verify a phone number or card, a Cloudflare Turnstile check runs in your browser. It sends Cloudflare your IP address and technical details of your browser and connection, so it can tell people from automated sign-ups. Cloudflare also uses this data on its own account to improve its bot detection, and for that use it is an independent controller (see Cloudflare's Turnstile privacy notice).

Email — Resend (transactional email delivery).

Device notifications — your browser's push service (Google, Apple, Mozilla, Microsoft or Samsung, depending on your browser). If you allow notifications in a browser, each notification is sent to your browser maker's push service, which delivers it to that browser. Your browser, not us, decides which service is used, and we have no agreement of the kind described above with these services; each operates under its browser maker's terms. The text of a notification is encrypted so that only your browser can open it: the service cannot read it and sees only your browser's notification address and technical details such as the size, priority and time of the notification. Once a notification reaches your device it may appear on your lock screen, depending on your device settings. If you turn on "Hide message previews" under Account → Notifications, notifications leave out the text of the message or prompt and show only who wrote or what finished.

Analytics — PostHog (EU cloud, Frankfurt), only with your consent.

Sign-in — Google (independent controller if you choose Google sign-in).

AI assistants you connect through the MCP connector (for example Anthropic's Claude or OpenAI's ChatGPT) receive the data you ask them to act on and are controlled by their own providers under their own terms. You decide what to connect and can revoke it at any time.

Team members. If you join a team, the team leader can see the content in the team workspace, your name and email, and the team's credit spend.

Legal and corporate. Authorities where the law requires; professional advisers; a buyer or successor if the business is transferred, who must honour this notice. If an authority or a court asks us for your data, we will tell you before we disclose it, unless the law prohibits it or telling you would put someone at risk or prejudice an investigation.

Our database and application run in the European Union (Germany). Some processors are in the United States: Atlas Cloud AI and the model providers it routes to, Anthropic, TypeSafe AI (message and image reading), Amazon Web Services (media storage and content checks), Cloudflare (bot protection) and Stripe. Prelude, our phone verification provider, is based in France and states that it hosts data on Amazon Web Services servers in the European Union; it also states that its own service providers may transfer data outside the European Union under the safeguards of Chapter V of the GDPR. The text message itself is carried by telecommunications operators, including the operator of your number. DeepSeek, which runs LuviBot’s language model, processes the content you send to LuviBot in China (see section 4). Device notifications are delivered by the push service of a company headquartered in the United States (Google, Apple, Microsoft, Mozilla) or South Korea (Samsung), depending on your browser (see section 4); the notification text is encrypted while it does so.

For transfers to the United States we rely, in this order:

• on the UK Extension to the EU–US Data Privacy Framework (the "UK–US Data Bridge") where the recipient is certified under it, which the UK Government has recognised as providing adequate protection; and
• otherwise on the recipient's data processing agreement incorporating the EU Standard Contractual Clauses with the UK International Data Transfer Addendum approved by the Information Commissioner, together with our assessment of the transfer.

Where a recipient holds a Data Bridge certification, its agreement with us also contains the Standard Contractual Clauses as a fallback. You can ask privacy@luvicreator.com which mechanism applies to a given recipient and for a copy of the relevant safeguard.

We may keep data longer where needed to establish, exercise or defend legal claims, or where the law requires.

You have the right to: access the personal data we hold about you; have inaccurate data corrected; have data erased; restrict or object to processing based on legitimate interests; receive the data you provided in a portable format; withdraw consent at any time where processing is based on consent (this does not affect processing before withdrawal); and not be subject to a decision based solely on automated processing that has legal or similarly significant effects on you (we do not close accounts by automated means alone; see section 8).

To exercise a right, email privacy@luvicreator.com from your account email. We will respond within one month, extendable by two months for complex requests, and may ask you to verify your identity. You can also download your Output from your library at any time, delete content yourself, and delete your whole account from the account page (deactivated at once, permanent after a 14-day grace period, after a double confirmation). You can switch notifications off by type, pause them, or turn them off for a browser under Account → Notifications; removing LUVI's notification permission in your browser's site settings also stops notifications to that browser.

You have the right to complain to the Information Commissioner's Office (ico.org.uk, 0303 123 1113). If you live in the EEA you may complain to your local supervisory authority. We would welcome the chance to address your concern first.

Model providers apply automated safety filters to prompts and generated content. We also run our own automated checks:

• an automated reading of every message you send to LuviBot and a policy check before LuviBot prepares a paid generation (see section 4); if either finds that a request is likely to break our Terms, LuviBot does not prepare it and tells you why;
• an automated reading of every image you attach to LuviBot, upload or generate, which notes what it shows (for example a product or a person, text, a plain background, the number of people) to place your files and sort your library into categories; it does not refuse content and decides nothing about you or your account (see section 4);
• a check of every uploaded image, every image used in a generation request and every generated image for nudity and sexual content, which also estimates the apparent age of any face shown (Amazon Rekognition, see section 4).

When a check refuses something, we tell you what was restricted, which rule of our Terms it breaks, that the decision was made by automated means, and how to appeal. The one exception is a refusal because content may sexualise a minor: then the notice may say only that the action could not be completed, without the reason, as giving it could hinder a report to the authorities. That notice still lets you tell us you think it is a mistake. When a check flags content without refusing it, we do not notify you; our staff review it and, if it is found unlawful, act as described above. A refused upload is not added to your library, and a request refused before it runs is not charged; a result refused after the model has run is withheld and charged, as set out in section 4.3 of the Terms. We do not close accounts by automated means alone: a person reviews the case first.

If you think something was blocked by mistake, use the "Blocked by mistake?" or "I think this is a mistake" option shown with the refusal or write to abuse@luvicreator.com. A person reviews it and we reply within 30 business days. These checks are not used to identify you or anyone else, and we do not profile you for advertising.

Data is encrypted in transit (TLS) and at rest. Access to production systems is restricted to staff who need it, protected by multi-factor authentication and logged. LUVI staff with administrator access can open your LuviBot conversations only to respond to a legal request, to investigate misuse or to handle a support case; each access requires a recorded reason and is logged. Media is served only through signed, short-lived credentials. We keep encrypted backups in a separate, write-protected store. No system is perfectly secure; if we discover a breach affecting your data we will notify you and the ICO as the law requires.

The Services are for adults. We do not knowingly process data of anyone under 18 and will delete it if we learn we have.

We will post changes here and update the date at the top. For material changes we will email you or show a notice in the Services before they take effect.

TVN Global Limited (trading as LUVI Creator), 82a James Carter Road, Mildenhall, Suffolk, IP28 7DE, United Kingdom

privacy@luvicreator.com · Data protection lead: the Data Protection Lead, reachable at the same address. We are not required to appoint a Data Protection Officer and have not appointed one.