Privacy Policy
This policy describes how Shohei Yoshida ("the operator"), who provides Goipon ("the Service"), handles user information. This is a reference translation; if it conflicts with the Japanese version, the Japanese version prevails.
Last updated: November 10, 2026
Information we collect
- Account information (email address and authentication information. Passwords are hashed by the authentication provider and are never available to the operator in plain text)
- If you sign in with Google, the information we receive from Google (email address, name, profile picture URL, and Google account ID). We do not receive other information in your Google account (such as contacts, Gmail, or Drive)
- Content you create or select (dictionary words and meanings, selected terms and nearby sentences, source information, study history, and projects)
- Terms you explicitly select with the browser extension, the surrounding sentence, page title, and source URL (for YouTube, as described under "Goipon browser extension" below)
- If you turn on browser notifications, the delivery information for them (the push endpoint URL and encryption keys issued by your browser, and a string identifying the browser type)
- Usage and technical data (generation count, AI processing volume, IP address, device and OS information, and error information that excludes personal data by default)
- Usage records for improving the Service (types and counts of actions such as opening the app, picking terms, and finishing reviews; a random per-device ID; at sign-up, the referring campaign name and referrer domain; and whether you opened the Service from an email or notification). These do not include the terms or text themselves. The pre-sign-up trial page (goipon.com/try) also records a random per-device ID and the types of actions
- Payment and subscription information (Stripe customer ID, plan, subscription status and renewal date, and when you agreed to the Terms when ordering. We do not receive card numbers)
- When your Pro trial started and when it ends
- Invite program records (invite code, which account invited which, and when and how many bonus words were granted). Inviters see only the number of people and words, not the invited people's email addresses or other details
- Feedback (what you send with "Send feedback" in Settings or Account, or in the one-time question shown after a review; the screen type you optionally choose, your device type, the app version, and when you sent it. We also record when the question was shown)
- Reports on AI output (the word, the meaning written by AI, and what you write, sent with "Report a mistake" on a word's page and elsewhere)
- Records of your consent to and withdrawal from AI processing (the time and the version of the explanation you agreed to)
Purpose of use
- Providing the Service (writing, saving, and reviewing meanings in your dictionary)
- Managing usage limits, the Pro trial, invite bonuses, and paid plans
- Addressing issues
- Aggregating and analyzing usage records and sign-up referral sources to improve features and measure the effect of advertising and invite programs (not for delivering personalized ads). We also record whether you opened the Service from an email or notification to improve delivery
- Review reminders (at most once a day, when you have terms due and have not reviewed yet that day, delivered as a browser notification (Web Push), an on-device notification on Android, or, if you don't receive notifications, an email to your registered address. You can turn them off at any time in Settings, in your browser or device settings, or from the link in the email)
- Weekly summary (an email sent once a week to your registered address with your study record, such as the number of well-learned words and words you recalled after a month. Depending on your settings, it includes words you recalled that week and the titles of their sources. You can stop it, or choose not to include words, at any time in Settings or from the link in the email)
- Handling feedback and reports on AI output (we read them to improve the Service and investigate issues, and reply to your registered email address when a reply is needed. We count answers and their shares to guide improvements. We do not publish what you write or send it to AI)
- Responding to violations of the Terms of Service or misuse
AI processing
With your explicit consent, data is sent to the API of Anthropic, PBC (United States) only in the following cases:
- When writing meanings: the terms you selected, their nearby sentences, and the learning and display languages
- When you tap a word in the reader: the tapped word, the one sentence containing it, and the language of the source (to show its meaning in that sentence). The result is stored on our server under a key computed as a hash of the word, sentence, and language; the sentence itself is not stored. Without consent, tapping a word shows only the built-in dictionary data (EJDict), and nothing is sent to AI.
The original PDF, images, full web pages, browsing history, source URLs, and page titles are not sent to AI. The browser extension runs only when you invoke it. By default, Anthropic does not use API inputs or outputs for model training and deletes them within 30 days. They may be retained longer for Usage Policy enforcement, legal requirements, or separate contractual terms. You can withdraw consent in Settings; no new AI transmission occurs until you consent again. Anthropic is located in a foreign country (the United States). The US system and Anthropic's measures are described under "Providing data to providers in foreign countries" below.
On the web, PDFs are displayed on your device. On mobile, a PDF may be uploaded temporarily for text extraction and is deleted after processing, although it may remain for a limited period after a processing failure or in backups.
When you open a web page from a URL, our server fetches the page, extracts the article text, and returns it to you. When you open a PDF from a URL on the web (including "Open this PDF in Goipon" in the browser extension), your browser fetches it directly from its host; only if the host does not allow this, our server fetches it and returns it to you. In both cases, the fetched content is not stored on our servers. When you add words to your dictionary, the URL is saved as their source.
Service providers
As needed, we entrust the handling of personal data to the following providers:
- Supabase (authentication and data storage)
- Anthropic (AI processing, only with your consent)
- Sentry (error monitoring configured not to send personal information by default)
- Stripe (payments)
- Resend (sending email. It receives the recipient address and the message. Review reminders contain the number of terms due; the weekly summary contains your study record and, depending on your settings, words and the titles of their sources. To notify the operator when feedback arrives, it also receives the feedback and the sender's email address)
- Cloudflare (website delivery, and forwarding email for our contact address, [email protected])
- Google (sign-in with Google, and receiving email for our contact address (Gmail))
Browser notifications are delivered through the push service operated by your browser's vendor (such as Google, Mozilla, Apple, or Microsoft); the notification content (such as today's word count) is encrypted, so the push service cannot read it. Card payment details are not stored on the operator's servers.
Providing data to providers in foreign countries
All of the providers above are US companies, and data may be handled in the United States and other countries.
- Supabase, Inc. (authentication and data storage. The database and files are stored in Japan (Tokyo))
- Anthropic, PBC (AI processing, only with your consent)
- Functional Software, Inc. (Sentry; error monitoring)
- Stripe, Inc. and its group companies (payments)
- Resend (sending email)
- Cloudflare, Inc. (website delivery and email forwarding)
- Google LLC (sign-in with Google, the operator's inbox (Gmail), and delivering browser notifications)
- Mozilla Corporation, Apple Inc., and Microsoft Corporation (delivering browser notifications, depending on your browser)
For the US system for protecting personal information, see the survey results published by the Personal Information Protection Commission of Japan (https://www.ppc.go.jp/personalinfo/legal/kaiseihogohou/#gaikoku). The United States has no comprehensive federal law equivalent to Japan's Act on the Protection of Personal Information; personal information is protected through enforcement by the Federal Trade Commission (FTC), sector-specific federal laws, and state laws (such as the California Consumer Privacy Act).
With providers other than Anthropic, we work under data processing agreements (including each provider's data processing addendum) that require them not to use personal data beyond the entrusted purposes, to take security measures, and to manage their subprocessors. Anthropic states that by default it does not use API inputs and outputs for model training and deletes them within 30 days, and the operator uses the API under Anthropic's data processing addendum. Details of the measures each provider takes are available on request.
Goipon browser extension
Only when you press "Add to Goipon", the extension saves the selected word, the sentence containing it, and the page title and URL to your Goipon account. If you turn on the YouTube feature, it likewise saves the word you click in the captions shown in your player, that caption sentence, the video title, and a URL with the timestamp (it never downloads or stores captions or videos). It does not collect or send page text, browsing history, or form inputs. Only when you press "Open this PDF in Goipon" in a tab showing a PDF, it passes the PDF's URL and the tab title to the Goipon reader (how the PDF is fetched is described under "AI processing" above). To show translations, it downloads dictionary data split by the first two letters from our server (goipon.com); the word you look up is not sent. To connect the extension, we issue a separate sign-in session for the extension and store it in your browser. You can end this session at any time with "Disconnect" in the extension popup. Saved words and sentences are sent to the AI processing described in this policy (Anthropic's API) when you create meanings in the app. Goipon's use of information received from the browser extension adheres to the Chrome Web Store User Data Policy, including the Limited Use requirements.
Data retention and deletion
Your account and data are retained until you delete them. You can delete your account at any time from the Account screen; doing so deletes your dictionary, sources, saved quotes, study history, feedback, and account information (feedback notices already delivered to the operator's inbox may remain, like other emails sent to us). Data contained in backups or diagnostic logs may be deleted after a certain period. Payment records that Stripe keeps as required by law are retained under Stripe's policies. If you cannot sign in, see https://goipon.com/en/legal/delete-account for how to delete your account.
Disclosure to third parties
Except where permitted by the Act on the Protection of Personal Information, such as when required by law, we do not provide personal data to third parties without your consent. Providing data to the providers above is entrustment within the purposes of use.
Requests for disclosure and other actions
You may request notification of the purposes of use, disclosure (including records of provision to third parties), correction, addition, or deletion, suspension of use or erasure, and suspension of provision to third parties. Contact [email protected] from your registered email address. To verify your identity, we check that the request comes from your registered address and may ask for additional confirmation. There is no fee. Disclosure is provided as electronic records by email. You can also view and delete much of your data yourself in the app.
Security measures
- Basic policy and handling rules: this policy is our basic policy, and we have set handling procedures for each stage of collection, use, storage, and deletion.
- Organizational and personnel measures: the operator, as the sole person responsible, handles personal data and regularly reviews how it is handled.
- Physical measures: devices used for operations require sign-in authentication, as a precaution against loss or theft.
- Technical measures: communications are encrypted, and the database uses row-level access control so that each user can read and write only their own data. Administrative keys are kept in non-public locations.
- Understanding the external environment: personal data is handled by entrusting it to US providers (see "Providing data to providers in foreign countries").
Your choices and responsibilities
AI consent is optional, but writing meanings is unavailable without it. You can still read sources, pick words, and see built-in translations. Do not submit another person's personal or confidential information without authority.
Changes to this policy
We may change this policy to reflect changes in laws or the Service. We will announce the updated policy in the Service, and we will ask for your consent again before changing the purposes of use or the scope of AI transmission in ways you could not reasonably expect.
Operator and contact (complaints)
Operator: Shohei Yoshida
Contact: [email protected]
The operator's address will be disclosed without undue delay upon request by the data subject.