Serifu AI
JP

Privacy Policy


Effective Date: June 19, 2026

Last updated: August 24, 2026


This Privacy Policy ("Policy") describes how we handle user information in the Serifu AI app and related services ("Service"). By using the Service, you agree to this Policy.


1. Scope


This Policy applies to the collection, use, storage, and disclosure of information obtained through the Service.


2. Information We Collect


The Service may collect the following information:

  • Account identification information
  • Technical information about usage and devices
  • Device identifier (for usage limit management)
  • Script data, original PDFs, and generated or recorded audio imported by users
  • Account, consent state, and lightweight practice logs required for the web app and cloud sync
  • Website and web app usage information, such as page views, clicks, scrolls, errors, and masked interaction records
  • Information provided through inquiries

    • 3. Purpose of Use


    We use collected information for the following purposes:

  • Providing, maintaining, and improving the Service
  • Delivering AI voice reading features
  • Delivering AI script structuring features (automatic extraction of character names, dialogue, and stage directions from text)
  • Providing web-based script editing, original PDF viewing, practice features, cross-device sync, and restoration
  • Managing usage limits and preventing misuse
  • Confirming billing state and providing paid features
  • Usage analysis, quality improvement, and issue investigation
  • Responding to user inquiries
  • Compliance with legal requirements

    • 4. Script Data Handling


    Script data imported into the Service is temporarily sent to servers for the following AI processing:


  • Script Structuring: Text data is sent to Anthropic's Claude AI to structure it into character names, dialogue, and stage directions
  • PDF/Image OCR: PDF and image data is sent to Google's Gemini AI for optical character recognition
  • Voice Generation: Dialogue text is sent to Google Cloud Text-to-Speech to generate read-aloud audio

    • Temporary processing data created for AI features is promptly deleted after processing is complete, except for the voice-generation diagnostic data described below.


    To improve voice-generation quality and investigate issues, we retain the input sent to Google Cloud Text-to-Speech (including dialogue text, voice, pronunciations, and related voice settings) and the generated audio in a server-only area that users cannot access directly for seven days from creation. Data older than seven days becomes eligible for deletion and is normally deleted within 24 hours. We do not use this diagnostic data for AI model training, advertising, or marketing.


    If you use the web app or cloud sync, script text, character names, original PDFs, generated or recorded audio, and lightweight practice logs are stored in the cloud based on your consent. This data is used for web editing, original PDF viewing, web practice, cross-device sync, and restoration. We do not view script content except when necessary to provide the Service, respond to incidents, or comply with legal requirements.


    None of the above AI services use the submitted data to train (improve) their AI models. Under each service's API terms, data submitted via commercial APIs is explicitly excluded from model training purposes.


    However, AI service providers may retain submitted data for safety monitoring, abuse prevention, or logging enabled by the operator. Standard retention differs by service and configuration: Anthropic API data is generally retained for no more than 30 days, while Gemini API logging, when enabled, can be configured for up to 55 days. Except for cases such as legal compliance or abuse investigations, data is marked for deletion after the applicable retention period. These logs are not used for AI model training.


    4-2. Confidentiality of Script Data


    We treat all script data imported by users (including unpublished works) as confidential information and commit to the following:


  • We will not disclose or provide script data content to third parties beyond the AI processing necessary for service delivery
  • We will not view or use script data for our own purposes (such as marketing or analytics)
  • The AI service providers specified in Section 4 (Google LLC, Anthropic PBC) also protect the confidentiality of submitted data under their respective API terms of service
  • Transmission of script data through sharing features is based on the user's own intent, and management of information shared with recipients is the user's responsibility

    • 5. Retention and Deletion


    We retain information only for as long as necessary to fulfill the purposes described above. Upon request, or when the purpose has been achieved, we delete or anonymize information in accordance with applicable laws. Voice-generation diagnostic data is deleted according to the period stated in Section 4. Scripts, original PDFs, audio data, and practice logs stored through the web app or cloud sync are deleted through account deletion or cloud data deletion procedures.


    5-2. Analytics and Session Replay


    We may use Vercel Analytics and PostHog to improve quality and investigate issues, including reviewing page views, clicks, scrolls, and similar interactions. In the web app, session replay starts only after cloud sync consent, and script text, character names, PDF display, input contents, audio data, signed URLs, and similar sensitive information are excluded from recording or masked.


    6. Third-Party Data Sharing


    The Service uses the following third-party services as necessary to provide the Service:


  • Google LLC (United States): PDF/image OCR processing (Gemini AI), voice generation (Google Cloud Text-to-Speech)
  • Google LLC / Firebase (United States): Account authentication, cloud sync, storage of original PDFs and audio data, Cloud Functions, other backend processing, and usage analytics
  • Anthropic PBC (United States): Text script structuring (Claude AI)
  • RevenueCat, Inc. (United States): Subscription state verification and paid feature delivery
  • Vercel Inc. (United States): Website hosting and analytics
  • PostHog Inc. (United States): Web app usage analytics and masked interaction records

    • Third parties that perform AI processing use the transmitted data solely for AI processing and do not use it for AI model training, advertising, or marketing purposes. We do not share your information with other third parties without your consent, except:

  • When required by law
  • When necessary to protect life, body, or property

    • 7. Your Rights


    You may request disclosure, correction, suspension, or deletion of your information. We will verify your identity and respond within a reasonable timeframe.


    8. Security Measures


    We implement organizational and technical security measures to prevent unauthorized access, loss, or damage to information.


    9. Policy Changes


    We may update this Policy as needed. Continued use of the Service after changes constitutes acceptance of the updated Policy.


    10. Contact


    For inquiries about this Policy:

    hiraly.dev@gmail.com