PRIVACY POLICY

Privacy Policy

Effective: September 24, 2026Last updated: November 14, 2026

Tetsuya Takayanagi (the "Operator") handles the information of the users of Fluent, the English shadowing service at https://fluent-en.com and its iPhone app (together, the "Service"), in accordance with this Privacy Policy (the "Policy").

The Japanese version of this Policy is the governing text. This English translation is provided for convenience; where the two differ, the Japanese version prevails.

Contents

1.Operator

The Service is operated by Tetsuya Takayanagi, who is responsible for the handling of users' personal information. The postal address and telephone number are disclosed without delay on request. Contact: support@fluent-en.com.

2.Information we collect

To provide the Service, the Operator collects the following information. What you register, type, record or upload is collected when you do so; results of AI processing and learning records when you use a feature; billing information when you subscribe; and usage information while you use the Service.

  • Account information: email address, password (stored as a hash; the Operator never keeps it in plain text), sign-in links and other credentials, and settings such as the display language.
  • User Content: the audio you upload (the region you cut out), your shadowing recordings, script text and tags.
  • Text you type: your messages to the AI coach (chat), your answers in the AI class and the placement test (such as writing tasks), and the topics you enter for the AI class.
  • Goals and plans: your answers to the AI coach's intake questions (goal, deadline, self-assessed level, time available, topics of interest, difficulties and so on), the materials you already have, your study plans and their changes, and coaching records.
  • Results of AI processing: transcriptions, per-word timings, scores for pronunciation, rhythm and intonation, coach comments, what the AI heard, Japanese translations, word explanations, and the scores and level estimates of the AI class and the placement test.
  • Records of conversations with the AI: transcripts of the AI class, the conversation test, the AI coach and the placement-test interview, and the reviews, record summaries, level estimates, results and usage time based on them. The audio itself is not stored.
  • Consent records: whether you agreed to the Terms and this Policy, to sending data to AI services, and to voice conversations with the AI; when you agreed or withdrew, on which screen, and to which version of the notice.
  • Vocabulary: the words and phrases you save, their context, and the generated meanings, parts of speech and example sentences.
  • Practice and learning history: the materials practised, practice time, number of takes, scores and timestamps, and the AI classes and placement tests you took.
  • Billing information: the status of your paid plan (including the plan and billing interval), your Stripe customer and subscription IDs, for App Store purchases the product, expiry and renewal status (through RevenueCat), and the billing period. Credit card numbers are held by Stripe, the payment provider, and never reach the Operator.
  • Analytics data: IP address, browser and device type, pages and screens viewed, referrer, and identifiers such as cookies or the iPhone app's per-installation identifier (see "Cookies and analytics").
  • Usage and device information: the screens viewed, the features used, timestamps, the kind of device (computer or phone, the iPhone app and its version), a random identifier per browser or device, if you turn on notifications, the iPhone's notification identifier (push token), and, if you installed the iPhone app from an Apple Search Ads ad, that ad's campaign and similar details. It does not include the content of what you type or say.
  • Enquiries and feedback: the name, email address and text you send by email, through the contact form or through the in-app feedback form, the page URL at the time, and browser information.
  • Invites and the share reward: the invite code you set, who invited whom, the rewards given, and the URLs of social media posts submitted for the share reward with the result of their check.

3.How we use it

The information is used for the following purposes.

  • Providing and operating the Service: account authentication, storing and playing materials and recordings, and showing your vocabulary and practice history.
  • Providing the AI features: conversations with the AI teacher and coach, transcribing recordings and writing AI analysis comments, Japanese translations, word explanations, summaries of reviews and records, scoring and level estimates in the AI class and the placement test, study plan suggestions, and searching learning resources and suggesting materials. See "Information sent by the AI features, and consent" for what is sent to external AI services.
  • Managing paid plan subscriptions, charging fees and managing the free trial.
  • Answering enquiries and feedback, sending important notices (changes to the Terms or fees, incident information and so on), and sending the study reminders you turn on.
  • Analysing usage, measuring the effect of advertising, investigating problems, improving features and developing new ones.
  • Preventing misuse (repeating the trial with several accounts, unauthorised access and so on) and responding to breaches of the Terms of Service.
  • Meeting legal obligations.

4.Third parties and processors

The Operator does not disclose users' personal information to third parties except where the law requires it, where it is necessary to protect a person's life, body or property and obtaining consent is difficult, or where the user has consented.

To provide the Service, the Operator uses the services of the providers listed below and entrusts them with information only to the extent necessary. All of them are located outside Japan (mainly in the United States) and handle the information under their own privacy policies and their contracts with the Operator. In accordance with the Act on the Protection of Personal Information, the Operator checks the legal framework of each provider's country and the measures it takes before entrusting data to it.

The AI processing of audio and text (ElevenLabs, OpenAI) is used under settings and contracts that keep the data sent from being used to train their models.

  • Supabase: stores account information, materials, recordings and learning records
  • Vercel: delivers the website and provides cookie-free aggregate statistics
  • Stripe: payments for the paid plan (card numbers never reach the Operator)
  • RevenueCat: managing the status of in-app purchases made in the iPhone app (App Store) — your user ID and purchase information; Apple processes the payment — and measuring installs from Apple Search Ads (see “Cookies and analytics”)
  • OpenAI (USA): voice conversations with the AI teacher and coach (your voice is sent in real time; see "Voice conversations with the AI"), the generation of text such as AI analysis comments, Japanese translations, word explanations, summaries of reviews and records, scoring and study plan suggestions, the AI coach's search of learning resources (text embeddings), and web searches for candidate materials. See "Information sent by the AI features, and consent" for what is sent
  • ElevenLabs (USA): transcribes audio (your shadowing recordings, your spoken answers in the placement test, the answers you record in the TOEIC speaking test form of the AI class, and the audio you upload to create a material on the web are sent) and generates the audio of preset materials
  • Expo (650 Industries, Inc., USA): delivers the iPhone app's study reminders if you turn on notifications (the notification identifier and the text of the notification are sent; it never contains consultation details)
  • Resend: email delivery that forwards contact form messages (name, email address and message) to the Operator's mailbox.
  • Google Analytics (Google LLC, United States): analytics for the website and the iPhone app (cookies on the website, a per-installation identifier in the iPhone app; see "Cookies and analytics")
  • PostHog (PostHog Inc., United States): analysis of how the website and the iPhone app are used, and recordings of interactions on some pages of the website and some screens of the iPhone app (session replay, only when it is used). A pseudonymous identifier is used; your name, email address and what you type or say are not sent (see "Cookies and analytics")

5.Information sent by the AI features, and consent

When you use an AI feature of the Service (AI analysis, Japanese translation, generating word meanings, the AI class, the conversation test, the AI coach, placement-test scoring and so on), the following information is sent to external AI services, only to the extent needed for that processing.

  • Sent to OpenAI (USA): your voice (in real time, during conversations with the AI teacher and coach), transcripts of your voice and the text you type (messages to the AI coach, answers such as writing tasks), and learning information (goals, level, learning records such as scores and the words you stumbled on, the words and context in your vocabulary, the English text of materials, and study plans). It is used for conversations with the AI teacher and coach, AI analysis comments, Japanese translations, word explanations, summaries of reviews and records, scoring and level estimates, study plan suggestions, and text embeddings that let the AI coach search the Operator's learning resources
  • OpenAI web search (when the AI coach looks for candidate books and other materials): only a search text built from the skills you want to learn, your purpose, an approximate level and your budget is sent; the text you type and your conversations are not
  • Sent to ElevenLabs (USA): recorded audio (your shadowing recordings, your spoken answers in the placement test, the answers you record in the TOEIC speaking test form of the AI class, and the audio you upload to create a material on the web). It is used for transcription

Your name and email address are not sent (what you yourself say or write in a conversation or text is sent as it is). Data is sent over encrypted connections (TLS).

Both providers' APIs are used under settings and contracts that keep what is sent from being used to train their AI models. Each provider handles the information under its API terms and privacy policy. OpenAI may retain the data it receives for up to 30 days to monitor for abuse.

Both providers are US companies, and the information sent is processed outside Japan, including in the United States. How the Operator handles entrusting data to providers abroad is described in "Third parties and processors".

In the iPhone app, before an AI feature sends any information, the app shows a notice ("Sending data for AI features") that lists what is sent, to whom and why, and sends nothing until you agree. The notice appears once after you sign in and, if you have not agreed, again right before an AI feature is used. If you do not agree, only the AI features are unavailable; recording, playback, materials, your saved vocabulary and so on keep working.

You can withdraw or give your consent again at any time in the iPhone app under Manage › Privacy and data. After withdrawal the AI features cannot be used in the iPhone app, and voice conversations with the AI cannot be used on the web either. On the web, consent is obtained through your agreement to the Terms and this Policy when you start the free trial or sign up for a paid plan and through the notice before a voice conversation with the AI; you can withdraw it by contacting support@fluent-en.com. Whether you agreed or withdrew, when, on which screen and to which version of the notice is recorded in your account.

6.Voice conversations with the AI

In the AI class, the conversation test and the AI coach's voice consultations, your voice is sent in real time to OpenAI's voice conversation service during the conversation, which produces the spoken replies of the AI teacher or coach and a transcript of the conversation.

Before a voice conversation, the Operator asks for your consent to sending your voice. On the web, it is asked when you start the free trial or sign up for a paid plan, together with your agreement to the Terms and this Policy (users who registered before that are asked before they first use these features). In the iPhone app, agreeing to the notice described in "Information sent by the AI features, and consent" includes voice conversations. Whether you consented, when, on which screen and to which version of the notice is recorded in your account. If the notice changes significantly, you are asked again. Without consent these voice features cannot be used.

You can withdraw your consent in the iPhone app under Manage › Privacy and data or by contacting support@fluent-en.com; after withdrawal the voice features can no longer be used.

The Operator keeps the transcripts of the conversations, the reviews, record summaries, level estimates and results made from them, and the usage time. The audio itself is not stored. In the speaking part of the placement test and in the TOEIC speaking test form of the AI class (your answer is recorded to a timer), the audio of your answers is sent to ElevenLabs to be transcribed for scoring; this audio is not stored either.

The transcripts of the AI coach's voice consultations can be set not to be kept and can be deleted in the iPhone app under Manage › Privacy and data. Other transcripts and results are kept until you delete your account.

OpenAI's API is used under terms by which the audio and text it receives are not used to train its AI models. OpenAI may retain the data it receives for up to 30 days to monitor for abuse.

7.Retention and deletion

User Content, AI results, vocabulary and practice history are kept until you delete them or delete your account. Materials and recordings can be deleted individually on the Service at any time.

Transcripts and results of conversations with the AI are kept until you delete your account (the transcripts of the AI coach's voice consultations are deleted once the record is made if you choose so in the settings, and can be deleted individually). The audio of the conversations is not stored. Consent records are kept until you delete your account.

When you delete your account, your account information, materials, recordings, analyses, vocabulary and practice history are deleted within a reasonable period, including removal from backups.

To prevent repeated use of the free trial, however, an irreversible identifier derived from your email address (a hash) and the start time of your free trial are kept after you delete your account. To prevent repeated use of the share reward, the URLs of submitted posts and the same hash are also kept after you delete your account.

Transaction records held by Stripe (invoices, payment history) are kept for the period required by tax and other laws.

Access logs and feedback are deleted or anonymised once they are no longer needed for operating the Service.

The record of in-app usage is deleted when you delete your account (records made before signing in are kept only with a random identifier and are not linked to any person).

8.Cookies and analytics

The Service uses cookies to keep you signed in and for security. If you disable these cookies, the features that require signing in will not work.

To improve the Service, the Operator uses Google Analytics, a web analytics tool provided by Google LLC. Google Analytics sets a per-browser identifier in cookies (_ga and cookies whose names start with _ga_, kept for up to two years) and collects, in a form that does not identify individuals, the pages you view, your interactions on the site (events such as signing up, creating a material or running an AI analysis), the referrer, the time of the visit, your browser and device type, and your approximate region. IP addresses are not logged or stored, as specified by Google Analytics 4 (anonymised). The data is processed on Google's servers (including in the United States) under Google's Privacy Policy (https://policies.google.com/privacy) and "How Google uses information from sites or apps that use our services" (https://policies.google.com/technologies/partner-sites).

If you do not want Google Analytics to collect this data, install the Google Analytics Opt-out Browser Add-on (https://tools.google.com/dlpage/gaoptout) or disable or delete cookies in your browser settings (note that the cookies needed to sign in will be disabled too).

The Operator also uses Vercel Web Analytics, provided by the hosting provider Vercel, to count page views and interaction events without cookies. Visitors are distinguished only by a hash generated from the request and discarded after 24 hours; no identifier that could identify an individual is stored.

To measure the effect of advertising, the website carries the ChatGPT ads measurement tag (OAIQ pixel) provided by OpenAI, L.L.C. The tag tells OpenAI whether a visit came from a ChatGPT ad and when an account sign-up, a free-trial start or a paid subscription happens (it may store an identifier in your browser). Your name, email address and learning content are not sent. OpenAI handles this information under its Privacy Policy (https://openai.com/policies/privacy-policy/).

The website also carries the Meta Pixel, an advertising measurement tag provided by Meta Platforms, Inc. It sets a per-browser identifier in cookies (such as _fbp) and tells Meta which pages you view and when an account sign-up, a free-trial start or a paid subscription (including the plan price) happens. Your name, email address and learning content are not sent. Meta handles this information under its Privacy Policy (https://www.facebook.com/privacy/policy/), and you can manage its use for ads in Meta's ad preferences (https://www.facebook.com/adpreferences/).

The website also carries the X Pixel, an advertising measurement tag provided by X Corp. It sets a per-browser identifier in cookies and tells X which pages you view. To measure X ads, when an account sign-up, a free-trial start or a paid subscription (including the plan price) happens on the website, our server also tells X about it together with the ad click id, your IP address, your browser type and an irreversible SHA-256 hash of your email address (the iPhone app sends nothing). Your name, your email address itself and your learning content are not sent. X handles this information under its Privacy Policy (https://x.com/privacy), and you can manage its use for ads in X's settings under Privacy and safety › Ads preferences.

On the top page and the sign-up / sign-in page (the pages before sign-in), the Operator uses Microsoft Clarity, provided by Microsoft Corporation, to improve the page. It sets a per-browser identifier in cookies and records interactions on the page such as clicks, scrolling and mouse movement, which are analysed as heatmaps and session replays. The contents of input fields are not recorded. It is not used on the screens after sign-in. Microsoft handles this information under its Privacy Statement (https://privacy.microsoft.com/privacystatement).

In the iPhone app, to measure the effect of advertising, when you install the app from an Apple Search Ads ad (ads shown in the App Store), our processor RevenueCat obtains that ad's details (such as the campaign, ad group and keyword) through Apple's AdServices framework and stores them with your user ID. This information is provided by Apple and is not used to track you across other companies' apps or websites.

The iPhone app (version 1.0.3 and later) also includes the Meta SDK, provided by Meta Platforms, Inc., to measure the effect of Meta ads (Facebook and Instagram). The Meta SDK tells Meta when the app is installed or opened and when an in-app purchase is made (the product and its price), together with device information such as the model, OS version and language, your IP address and an identifier assigned to each installation. The app does not ask for permission to track you and does not send the advertising identifier (IDFA). Your name, email address, user ID and learning content are not sent. Meta handles this information under its Privacy Policy (https://www.facebook.com/privacy/policy/).

The iPhone app (version 1.0.5 and later) also includes Google Analytics for Firebase, provided by Google LLC, to improve the Service and to understand app installs and usage. It tells Google about first launches, launches and time in the app, the screens viewed (including the sign-up and plan screens), sign-ins and account creation (and the method, such as password, Apple or Google), and when a free trial starts or an in-app purchase is started or completed (the plan and billing period; depending on how the purchase is made, also the product and its price), together with the app and OS versions, the device model, the language, an approximate region estimated from your IP address and an identifier assigned to each installation. The app does not ask for permission to track you and does not send the advertising identifier (IDFA), and the Operator has not set this data to be used for Google's advertising. Your name, email address, user ID and the text or audio you enter are not sent. Google processes this information on its servers (including in the United States) under the Google Privacy Policy (https://policies.google.com/privacy) and "How Google uses information from sites or apps that use our services" (https://policies.google.com/technologies/partner-sites). The per-installation identifier changes when you delete the app. The website opt-out add-on does not apply to the iPhone app.

In addition, to improve the Service, the Operator records how the web version and the app are used (the screens viewed, the features used and the kind of device), together with your account and a random identifier assigned to and stored on each browser or device, and stores this in the Operator's own database. It does not include the content of what you type or say.

The Operator also uses PostHog, an analytics tool provided by PostHog Inc. (United States), to improve the Service. Our server sends PostHog the same usage information that is recorded in the Operator's own database above (the screens viewed, the features used, the kind of device, the app version and the time). Usage while signed in carries a pseudonymous identifier derived from your user ID that cannot be turned back into it; usage before signing in carries the random identifier of the browser or device (the identifier sent to PostHog changes to a new one when you sign out). Your name, email address, the user ID itself, what you type or say, the text of materials, the content of AI analyses and any location estimated from your IP address are not sent. On the website, for some or all browsers (the Operator decides the share), interactions on the page (clicks, scrolling and moving between pages) may be recorded in a form that can be replayed (session replay) and sent to PostHog. In these recordings all text on the page and all input fields are masked, audio and video are not recorded, and the sign-in and sign-up, plan and payment, settings, AI teacher and AI coach conversation and material creation pages are not recorded at all. In the iPhone app (version 1.0.15 and later), only the screens before signing in (the first screen, sign-in and sign-up, the free-trial conditions, Try Fluent, and the screens for browsing materials, listening, articles and the like while trying the app without an account) the guide shown right after your first sign-in, and, while signed in, the screens for browsing materials and the like (Explore, listening, the AI News list, your vocabulary book, and Level check and Class in the AI tab) are recorded this way (what the screen shows, as images, and where you tap) and sent to PostHog, and the app includes PostHog software for this purpose. Text on these screens, such as buttons and explanations, is recorded (in the vocabulary book, this includes the words you saved and their meanings), but input fields (email address, password, codes, search boxes, playlist names, the topic of an AI class and the like) and the AI analysis results of Try Fluent are masked, and nothing is recorded while the keyboard is shown. Session replays while signed in carry the pseudonymous identifier derived from your user ID described above. Audio is not recorded, and the other screens after signing in (your own materials on the home screen, practice and its results, AI teacher and AI coach conversations and chats, your study records, plans and purchases, settings and so on) are not recorded. The app's PostHog software sends nothing other than these session replays. Session replays are sent from the browser or the iPhone app directly to PostHog, so PostHog necessarily receives your IP address with them; PostHog is set not to estimate a location from it and not to store it. Information sent to PostHog is processed on PostHog's servers (in the United States or the EU) and handled under PostHog's privacy policy (https://posthog.com/privacy). Session replays are deleted after about 30 days and other usage information after about one year. If you delete your account, the PostHog records remain until then, but as they carry only a pseudonymous identifier they are not linked to you. To stop sending information to PostHog, turn off "Help improve Fluent" at the end of this section; it works without an account and without signing in. While you are signed in, the same setting is also under Settings › Data controls on the website. A browser that sends Global Privacy Control (GPC) is treated as if it were turned off. Usage in that browser, its session replays and what our server records for actions taken in it (such as AI analyses and a trial start) are then not sent to PostHog. The choice is stored in each browser (clearing the browser's site data undoes it). While you are signed in on that browser, it is also recorded on your account, and nothing about your account is then sent to PostHog, including usage in other browsers and in the iPhone app while you are signed in, and what our server records (such as AI analyses, the days you studied, and your trial and subscription status). This takes effect at once in that browser (its other open tabs included); for your account, what our server sends stops within about a minute, and in other browsers it may take a few minutes. Turning it on again while signed in also clears it for your account (using the Service signed in on a browser that still opts out or sends GPC records it again). In the iPhone app, the same account setting is available while you are signed in, under Manage › Privacy and data ("Help improve Fluent"; shared with the website's setting). If your account is set not to be sent, usage of the iPhone app while signed in and its session replays (the guide shown right after your first sign-in and the screens for browsing materials and the like) are not sent to PostHog. Usage and session replays of the iPhone app while signed out (such as the sign-in and sign-up screens, Try Fluent, trying the app without an account, and the AI analyses of Try Fluent) cannot be stopped and are sent to PostHog with the random identifier of the device. This setting and GPC stop only what the Operator sends to PostHog; they do not apply to the other tools in this section (such as Google Analytics, Vercel Web Analytics, Microsoft Clarity, the ad measurement tags, and the Meta SDK and Google Analytics for Firebase in the iPhone app). How to stop Google Analytics, and how to manage the use for ads at Meta and X, are described in their own paragraphs. The record in the Operator's own database continues.

Data controls

Turn this off to stop sharing usage data from this browser with PostHog. To also stop sharing from the iPhone app while signed in, sign in to the same account on the website and turn it off.

Fluent’s own usage records continue. This setting and GPC stop only what is sent to PostHog. They do not apply to Google Analytics, Microsoft Clarity, the ad measurement tags or the other tools (see “Cookies and analytics” in the Privacy Policy).

9.Security

To prevent the leakage, loss or damage of the information it holds, the Operator takes necessary and appropriate safeguards: encrypted connections (TLS), access control on stored data (designed so that only you can access your own data), hashed credentials, and the careful selection and supervision of processors.

10.Access, correction and deletion requests

You may ask the Operator to disclose, correct, add to, delete or stop using the personal information it holds about you. Send your request from your registered email address to support@fluent-en.com. After verifying your identity, the Operator responds without delay within the scope set by law.

Some information, such as your email address and display language, can be changed yourself in the Service's settings. Materials and recordings can be deleted on the Service. Your consent to sending data for AI features can be changed in the iPhone app under Manage › Privacy and data.

11.Changes to this Policy

The Operator may revise this Policy in response to changes in the law or in the Service. Significant changes are announced on the Service or by email to the registered address. The revised Policy applies from the moment it is published on the Service.

12.Effective date

This Policy applies from September 24, 2026.

Revised on September 26, 2026 (operator name, processors).

Revised on September 29, 2026 (voice conversations with the AI and the consent to them added; what is sent to OpenAI and ElevenLabs corrected).

Revised on October 11, 2026 (what is kept to prevent repeated free trials; the record of in-app usage; the ChatGPT ads and Meta measurement tags).

Revised on October 12, 2026 (information collected for invites and the share reward; the contact form and the processor Resend; Microsoft Clarity on the top page).

Revised on October 13, 2026 (the X measurement tag; Microsoft Clarity on the sign-up / sign-in page).

Revised on October 16, 2026 (what the AI features send to OpenAI and ElevenLabs and why, the iPhone app's consent before sending and how to change it; the information collected, the purposes and the processor Expo added).

Revised on October 20, 2026 (Apple Search Ads measurement in the iPhone app and the processor RevenueCat's role in it added).

Revised on October 23, 2026 (Meta ads measurement in the iPhone app (the Meta SDK) added).

Revised on October 25, 2026 (Google Analytics for Firebase in the iPhone app added).

Revised on October 30, 2026 (the answers recorded in the TOEIC speaking test form of the AI class are sent to ElevenLabs for transcription).

Revised on November 7, 2026 (the PostHog analytics tool, session replay on the website and the setting that stops sending to PostHog — on this Policy's page and under Settings › Usage analytics; for the whole account while signed in — added).

Revised on November 9, 2026 (session replay in the iPhone app on the screens before signing in and the guide right after the first sign-in, and the app's Help improve Fluent setting under Manage › Privacy and data, added).

Revised on November 13, 2026 (the browsing screens while trying the iPhone app without an account added to its session replay).

Revised on November 14, 2026 (the screens for browsing materials and the like while signed in — Explore, listening, the AI News list, the vocabulary book, and Level check and Class in the AI tab — added to the iPhone app's session replay).