Privacy Policy

This Privacy Policy explains how Copycat Cafe Limited (“Copycat Cafe”, “we”, “us”) collects, uses, shares, and protects personal data when you use our website, app, lessons, AI and speech features, subscriptions, emails, and support. It is written for the EU General Data Protection Regulation (GDPR) and similar laws. The summary below gives the main points; the numbered sections give the detail.

1. Who we are

Copycat Cafe Limited is the controller of your personal data. We decide why and how it is used.

2. Data we collect

• Account data: name, email address, login details, settings, language and communication preferences, invitations you send or receive, and details of promotional access you have been granted.
• Subscription and billing data: plan, status, renewals, cancellations, refunds, transactions, and Paddle customer and subscription identifiers. Paddle handles your card and payment details; we do not store full card numbers.
• Learning data: lessons watched, exercises, answers, translations, flashcards, progress, level, languages, lesson searches, chat messages, corrections, scores, and feedback.
• Voice and speech data: Copy phrase recordings, live microphone audio while you use a speech feature, transcripts, pronunciation scores and analysis, target phrases, and audio we generate for you.
• Recording diagnostics: measurements such as recording length, volume, clipping, silence, background noise, audio format, microphone settings, and browser, operating-system, and device type. These do not include the recording itself, your full user-agent string, or a persistent device identifier.
• Support and contact data: messages you send us, your support history, and testimonials you choose to submit.
• Technical and usage data: IP address, browser and device information, logs, error reports, cookies and session identifiers, security signals, analytics events, and referral or affiliate information.

Public share links: if you choose to create a public result link, the page shows what you confirmed before publishing: a display name (your first name by default, or “A learner” if you remove it), your result, the copied phrase or conversation highlight, the language, and any share image. It never shows your recording or your private chat transcript. Anyone with the link can view, reshare, screenshot, or cache it. New links expire after 30 days. In Privacy & sharing in your account settings, you can change the display name, revoke a link immediately, or delete the result and its image. We cannot remove copies that others have already saved or reposted.

3. Why we use data and our legal bases

GDPR requires a legal basis for each use of personal data. The table groups our purposes by the basis we rely on.

Microphone access is a browser permission, not a legal basis. When you allow it and use a speech feature, we process your voice to provide that feature under our contract with you. You can decline or revoke microphone access in your browser at any time.

4. Voice, recordings, and AI features

When you record a Copy phrase

1. Your browser sends the recording to Copycat Cafe.
2. We store the original take in private storage, linked to that pronunciation attempt, so you can replay it later.
3. We send the audio, the target phrase, and your language settings to Langcraft to score your pronunciation. Langcraft processes the audio and transcript only while scoring and does not keep them afterwards. It keeps limited operational records such as request ID, status, and time.
4. We keep the result with your pronunciation history: the score, transcript, word-level analysis and feedback (Langcraft’s response with any audio removed), the target phrase, and recording diagnostics.

Only you can replay your recordings. They are not shown to family, team, or school plan managers, and they never appear on public share pages. We keep them while your account exists. You can ask us to delete specific recordings at any time. When you delete your account, they are removed from live storage, normally within a few hours. Encrypted backup copies are removed on a rolling schedule, normally within about 90 days.

Live speech features

For speech-to-text, your browser streams microphone audio to Soniox using a short-lived credential we issue, and Soniox returns a transcript. When you choose Call with Copy, your browser sends live microphone audio directly to Google Gemini Live using a short-lived, one-use credential. Google receives the audio, live transcripts, and the limited chat or lesson context and settings needed to hold the conversation, and returns Copy’s live transcript and generated voice.

We do not store your microphone audio or Copy’s audio from these live features. We save finalized text exchanges shown in chat with your ordinary chat history. Call with Copy uses Google’s session-resumption feature so an interrupted call can reconnect; Google says this can retain the conversation state, including text and audio, for up to 24 hours. Google may also keep prompts and responses for a limited period for abuse monitoring and legal obligations. Under the Gemini API terms that apply to our use, Google does not use this content to improve its products.

AI chat, lesson search, and generated audio

To answer a chat message, correct or grade your work, or evaluate a practice session, we send an AI model the relevant text, chat and lesson context, transcripts, and settings. To prepare lesson search, we send each authored lesson’s title, level, practice goals, English dialogue, and learning-language dialogue through OpenRouter to Google Gemini Embedding 2. We store the resulting numerical lesson vectors in our Render PostgreSQL database. When you search, we send only your search words to the same model and compare the returned query vector with the stored lesson vectors. We do not store the query vector. We do not send your account ID, email address, progress, chat history, or analytics history for this search. To read Copy’s replies aloud, we send the text to be spoken and voice settings to a text-to-speech provider. We store the generated audio privately so you can replay it. Text-to-speech providers never receive your own recordings. Section 8 lists each provider.

No AI training on your content

We do not use your chats, recordings, or transcripts to train AI models. We only use AI and speech providers whose terms, or our account settings with them, exclude our customers’ content from training or improving their models unless we opt in. We have not opted in, and we will not do so without updating this policy and obtaining your consent where the law requires it.

This is not a promise that providers keep no data. Some process your content only while handling the request (for example, xAI under the zero-data-retention setting we have enabled, and Langcraft). Others may keep limited data for a period to detect abuse, maintain security, or meet legal obligations. Section 8 describes what we know about each provider.

Human review

Authorized Copycat Cafe staff and contractors may review chats, transcripts, recordings, scores, logs, and screenshots when needed to help you, investigate security or abuse, fix bugs, check quality, improve prompts, or evaluate AI and speech models. Access is limited to people who need it for their work, and they are bound by confidentiality. Our engineers sometimes use Amp, an AI coding assistant, for this work. When they do, they include only the minimum excerpts needed. We treat this review as running and improving the service, not as training an AI model.

No voice identification or automated decisions

We do not use your voice to identify you, verify who you are, or create a voiceprint or other biometric identifier. Your recordings are used only for pronunciation feedback, replay, support, and quality review.

We do not make decisions about you based solely on automated processing that have legal or similarly significant effects. AI scores, corrections, and feedback are learning aids. Paddle and other payment providers may use automated checks for fraud prevention, payment authorization, and tax under their own privacy notices.

You can use Copycat Cafe without your voice. You can watch and read lessons, review content, and type in chat without granting microphone access. Pronunciation scoring, speech-to-text, and spoken practice need your audio, because we cannot score or transcribe speech we never receive.

Public microphone check: the check at /microphone-check analyzes and plays your sample entirely in your browser. Neither the sample nor its measurements are sent to Copycat Cafe, and closing, refreshing, or resetting the page discards them.

5. Sensitive information

We do not ask for special categories of personal data, such as information about health, racial or ethnic origin, religious or political beliefs, or sexual orientation, and ordinary language practice does not need them. Free-form chats, recordings, searches, and support messages can reveal them if you choose to share them, and an accent can suggest where someone comes from. Please leave out sensitive details that your request does not need.

We do not use your content to infer or categorize sensitive characteristics. If sensitive information appears incidentally, we limit its use to providing the feature, responding to you, keeping the service secure, and deleting or restricting it when it is no longer needed. If a request genuinely requires us to use special-category data, we will identify an applicable condition under GDPR Article 9, such as your explicit consent or the establishment, exercise, or defence of legal claims. You can delete the content or ask us to delete it at any time.

6. Analytics and measurement

We measure how Copycat Cafe is used so we can fix what does not work and improve lessons, search, signup, and checkout. We do this in two ways: with our own first-party analytics, stored in our database, and with Plausible’s cookieless website analytics. Neither is used for advertising.

Your choices: you can opt out in Data & privacy in your account settings. We also stop general analytics when your browser sends a Global Privacy Control or Do Not Track signal, and we exclude traffic we detect as automated. An opt-out does not affect records we need to run the service, such as sign-in, billing, and security events, or affiliate attribution needed to pay earned commissions.

First-party analytics

Unless analytics is off for you, we record visits and a fixed list of product events. A visit uses a random identifier held in our encrypted session cookie. The identifier changes every day (UTC) and a visit ends after 30 minutes of inactivity, so we cannot follow an anonymous visitor from one day to the next. A visit can include the landing page, the referring site and page without query strings, campaign labels (UTM parameters), broad browser, operating-system, and device type, and an approximate location (country, region, and city) that Cloudflare derives from your IP address. We do not store your IP address or full user-agent string in these records.

Events describe actions such as starting checkout, completing a lesson, or following a link to another site. For links to other sites, we record the destination site and page, with query strings and fragments removed. We never put private invitation links, access tokens, recipient contact details, or message content into analytics. If you sign in or create an account, we link the current visit and its events to your account, so we can see whether visits and signup steps lead to subscriptions.

Lesson searches: when you are signed in and analytics is on, we record each lesson search: the words you typed, the language, the level and progress filters, how many lessons matched, and which lessons were shown. We use these records to improve search results and the lesson catalog. We delete them after 45 days. They stay in our own database and are not sent to Plausible or any advertising service. Because search text is free text, please do not type personal details into lesson search.

Acquisition: when you create an account with analytics on, we keep one short record of how you first arrived: the landing page, the referring site without its page or query, campaign labels, whether the visit was direct or came from a recognized paid ad click, and the time. We do not keep advertising click IDs, full referrer addresses, IP addresses, or user agents in this record.

Email links: links in our service and lifecycle emails can carry a random delivery code that works for 30 days. If analytics is on and you follow such a link, we can connect that email to the resulting visit and later app activity. We remove the code from the address bar before the page loads. It never signs you in or grants access. If the visit is later signed in by a different person, we remove the connection. Email opens and clicks reported by our email provider are kept separate and are not treated as app use.

Signup-flow tests: when you create an account, we randomly assign it to one of two versions of the pages that follow signup, such as which plan page you see first. The assignment is saved with your account so every device shows the same version. It never changes prices, trial length, or access terms, and it sets no extra cookie. If analytics was on when you signed up, we compare aggregate checkout, trial, payment, cancellation, and refund outcomes between the versions. This is not an advertising identifier and is not sent to Plausible. A later opt-out stops the measurement but keeps your version. Deleting your account removes the assignment.

Plausible

Plausible gives us aggregate website statistics. It receives page addresses (with private or tokenized addresses replaced by a generic pattern), time on page and scroll depth, campaign labels, the referring site and page without query strings, device and browser type, approximate location, and a few conversion goals. It does not use cookies or other stored identifiers, does not store IP addresses or full user agents, and resets its anonymous visitor count every day. We do not send Plausible account, customer, or transaction IDs, affiliate or referral codes, form contents, download paths, or the destinations of outbound links.

7. Cookies and similar technologies

• Necessary: an encrypted first-party session cookie for sign-in, security, signup and checkout continuity, preferences, and subscription access. We also use limited browser storage for security, interface state, checkout, and the installed app. The same session cookie carries the daily analytics identifier described in section 6 when analytics is on.
• Affiliate referrals: if you arrive through an affiliate link, Affonso stores a referral ID in the first-party affonso_referral cookie for up to 60 days and may add an affonso_id parameter to the address. We also keep the referral in our session. If you sign up, we save a copy of the referral with your account as a fallback. At checkout, the current Affonso cookie takes precedence over that copy, and we pass the referral to Affonso and Paddle so the partner is credited.
• Operational and billing tools: Sentry collects minimized error and performance diagnostics. Paddle Retain (ProfitWell) runs only for signed-in users with a Paddle customer record, to show payment-recovery and subscription notices.
• Embedded YouTube videos: blog articles show a preview served by us. YouTube is contacted only after you choose “Accept YouTube and play video.” YouTube may then use cookies and similar technologies under Google’s Privacy Policy. We do not remember this choice, so each video needs its own activation.
• No advertising trackers: we do not use Meta Pixel, the Meta Conversions API, retargeting pixels, or cross-site advertising trackers.

8. Who receives your data

We share personal data only with the providers we need to run Copycat Cafe. Most act as our processors: they may use the data only on our instructions and under data-processing terms. Paddle, as merchant of record, and YouTube act as independent controllers for their own purposes, so their privacy notices also apply. We keep an internal register of these providers and update this list when we add one or change what one receives.

We previously used Resend to deliver email. It no longer receives new data from us, but it may retain records of earlier deliveries under its terms. Our own records may also contain delivery metadata from former providers.

We may also disclose data to professional advisers, to authorities when the law requires it, or to a buyer if our business is sold, under appropriate confidentiality.

9. International transfers

We are based in Cyprus, in the European Economic Area (EEA). Several providers in section 8 process data outside the EEA, in particular in the United States, and some also in the United Kingdom, Australia, or other countries. AI and speech processing can take place in different locations depending on the model and provider, so we do not offer EEA-only processing.

When personal data leaves the EEA, we rely on a European Commission adequacy decision, such as the UK decision or the EU-U.S. Data Privacy Framework for certified recipients, or on the Commission’s Standard Contractual Clauses included in the provider’s data-processing terms. We add measures such as encryption in transit and sending providers only the data a feature needs. You can ask us for information about the safeguard used for a particular provider.

10. How long we keep data

We keep personal data only for as long as we need it for the purposes above. The main periods are:

Longer retention can apply when data must be preserved for security, fraud prevention, a dispute, or a legal obligation. Providers keep data according to the terms described in section 8.

11. Deleting your account

You can delete your account from your profile. Some situations must be resolved first:

• If an account you own has an active subscription or trial, the deletion flow asks you to confirm separately. It then cancels billing immediately, ends access, and deletes the account. Deletion does not trigger a refund, so contact support first if you want to use the money-back guarantee. If automatic cancellation is not available, cancel from Billing first.
• If an account you own includes other learners, transfer or remove that account first, so that deleting your profile does not remove someone else’s access or records.

Deletion removes your profile, learning activity, chat history, recordings, generated audio, public result links and share images, and analytics records linked to you. We keep limited billing, promotional, and operational records, detached from you or anonymized where possible, when they are needed for accounting, security, fraud prevention, support, or legal claims. Paddle and other independent controllers keep their own records under their legal obligations.

Copies in backups, security logs, and provider systems are not removed instantly. They are deleted on their normal schedules unless they must be preserved for security, payment, tax, dispute, or legal reasons.

12. Security

We protect personal data with technical and organizational measures, including HTTPS, encrypted file backups, access controls, authentication, monitoring, restricted staff access, and security requirements for our providers. No online service can guarantee perfect security.

13. Family, team, and school plans

If you join through a family, team, school, or similar plan, the person or organization managing it can see limited information about you: your email address, invitation and seat status, subscription access, practice dates, lesson titles, Watch, Copy, and Chat completion, pronunciation score history, and the phrases those scores relate to. They cannot see your chats, transcripts, recordings, answers, or feedback.

If we offer additional school or team dashboards, we will describe what administrators can see before you sign up for that plan.

14. Your rights

Under GDPR, you have the right to:

• access your personal data and receive a copy;
• have inaccurate data corrected;
• have your data deleted;
• restrict how we use your data;
• receive data you gave us in a portable format;
• object to uses based on legitimate interests, such as analytics and product improvement, and to direct marketing at any time;
• withdraw consent at any time, without affecting earlier processing.

You can manage analytics in Data & privacy, public links in Privacy & sharing, and marketing emails through the unsubscribe link in each email. For anything else, contact [email protected]. We may need to verify your identity, and we will respond within one month, or tell you if we need more time as the law allows. Some data may still be needed to provide the service, keep it secure, meet legal obligations, process payments, prevent fraud, pay earned commissions, or resolve disputes.

You can also complain to a data protection authority. In Cyprus, this is the Office of the Commissioner for Personal Data Protection. You can also contact the authority where you live or work.

15. Children

You must be at least 16 years old to use Copycat Cafe. We do not knowingly collect personal data from children under 16. If you believe a child under 16 has used the service, contact us so we can review and delete the data where appropriate.

16. Changes to this policy

We review this policy when we add a provider, feature, or cookie, or change how we use or keep personal data, and we update it before or alongside the change. The date at the top shows the latest version. If a change is material, we will take reasonable steps to tell you and, where required, ask for your consent.