Scribo Privacy Policy
Languages: This Policy is available in Simplified Chinese, Traditional Chinese, English, Japanese, and Korean. All language versions are equally authentic. If the version in the language you use differs from another version, the version in the language you use prevails for that use. No version limits any mandatory personal-information-protection rights available to you under the laws of your habitual residence. The Simplified Chinese version is the maintenance source; to the extent not covered by the preceding rules, the Simplified Chinese version prevails in the event of any inconsistency.
This Policy explains how the Scribo App, the Scribo API, and feedback channels process information. The personal information processor is the Scribo team (“we”). We follow the principles of lawfulness, legitimacy, necessity, openness, transparency, and security protection. For privacy questions, complaints, or personal information rights requests, contact us through “Contact Us” in the App or by email at yzx_ios@163.com. Except for normal network processing necessary to return public startup configuration and legal pages, third-party components that collect or report information are enabled only after you actively agree to this Policy.
1. Core Processing Boundaries
- Recordings, imported audio, transcripts, marks, titles, search content, and the current meeting’s speaker features are mainly processed and stored locally on the device. Post-recording transcription, punctuation processing, and speaker separation within a single meeting are performed locally on the device. Only temporary or de-identified numbers for the current meeting are retained; real identities are not identified, and no cross-meeting voiceprint database is created. The Scribo API does not receive this meeting content and does not create a cloud meeting content library.
- The Service currently does not provide cloud translation, AI-powered summaries, or real-time transcription, and does not use third-party AI inference services. The Android and HarmonyOS clients support importing audio or video files from the system file picker or other apps. The Android client first copies an imported file to the local cache, validates its format, size, and duration, and saves it to local installation space after you confirm the title and processing authorization. The HarmonyOS client copies the selected file to the App’s local installation space and starts on-device transcription. The audio is not uploaded to the Scribo API or third-party AI inference services. The iOS client currently does not provide an external audio import entry. Before future features change data flows, we will update this Policy and complete the corresponding notice and consent requirements in accordance with law.
- Model hosting services are used only to provide local model files and do not participate in recording recognition, speaker separation, or transcript generation. Depending on region, model, and dependencies, model files may be downloaded from ModelScope, Hugging Face, or GitHub. Before the first download, the App will explain the network information that may be processed and request your consent on the page.
- Voice, recordings, audio imported from other apps, and temporary features used to separate speakers within the current meeting may involve sensitive personal information. The relevant processing is used only to complete the recording, transcription, and current-meeting speaker separation that you actively initiate. Refusing microphone permission does not affect viewing existing content, but the recording feature will be unavailable. Before recording, you should explain the necessity and impact to the relevant persons and obtain the applicable authorization, consent, or approval.
- The Service currently does not enable install attribution, advertising attribution, Apple Search Ads attribution, SKAdNetwork attribution, Install Referrer attribution, or referral reward flows; the production client does not contain the AppsFlyer SDK or OneLink entry points.
2. Information We Process
| Information | Purpose and Processing Boundaries |
|---|---|
| IP address, time, URL, User-Agent, status, and security logs of public page requests | Used to return startup configuration and agreements, maintain connections, and prevent attacks; no installation identifier is actively attached before consent. |
| Random installation identifier, device token, system and App version, device model, network and request metadata | After your consent, used to establish a sign-in-free device session, ensure API compatibility, provide security protection, and diagnose faults. The token is stored in plaintext in the device’s system secure storage; the server stores the necessary information for verification. |
| Entitlement, quota, and transaction metadata | Used to display entitlements, verify purchases, restore purchases, process renewals and refunds, prevent duplicate grants, and perform financial reconciliation; it does not include recordings, transcripts, or payment passwords. |
| Feedback content, optional contact information, and necessary environment information | Used to handle issues, contact you, and diagnose faults. Please do not submit recordings, complete transcripts, passwords, or device tokens. |
| Local recordings, imported audio, transcripts, marks, titles, search content, and temporary speaker features | Processed in the App’s installation space on the current device. Temporary speaker features are not used for cross-meeting identification and are not uploaded to the Scribo API. |
| Model download network information | When downloading models from ModelScope, Hugging Face, or GitHub depending on region, model, and dependencies, the relevant services may process IP addresses, request time, User-Agent, file URLs, resumable-download Range headers, and standard secure connection metadata. Requests do not contain recordings, transcripts, meeting titles, search content, speaker information, or local file paths. |
3. Analytics, Crash Reporting, and Third-Party Services
- After your consent, Android and iOS classify the runtime region and initialize only one analytics and crash-analysis service: Umeng for the mainland China category and Firebase for other regions. HarmonyOS NEXT currently does not initialize Umeng or Firebase analytics or crash-analysis services. Services that are not selected are not initialized.
- The runtime region classification refers in order to a valid local reliable classification, a trusted network region classification, the iOS App Store Storefront or the Android app install source; when these signals are unavailable, the device system language may be referenced. This classification is not equivalent to your nationality, habitual residence, or real-time physical location; VPNs, proxies, roaming, store region, install source, or system language may cause the classification to differ from your actual location. Reliable classification results are stored locally on the device for up to seven days, and only the region category, source category, and validity period are stored; the raw IP, country code, Storefront, or install source used for classification is not stored. The processing boundary for raw IP as normal HTTPS connection information is described in Section 2 of this Policy.
- The Umeng Mobile Analytics SDK (U-App) and Umeng Application Performance Monitoring SDK (U-APM) are provided by Umeng Tongxin (Beijing) Technology Co., Ltd.; privacy rules: https://www.umeng.com/page/policy. On iOS, Umeng Apple Search Ads, SKAdNetwork, in-app collection, and overseas domain reporting are explicitly disabled, retaining only whitelisted analytics, crash, jank, and launch performance capabilities after consent; the partner for the iOS U-APM Zhuomuniao SDK is Guangzhou Dongjing Computer Technology Co., Ltd.; rules: https://yueying-docs.effirst.com/privacy.html.
- Firebase Analytics and Firebase Crashlytics are provided by Google LLC; privacy information: https://firebase.google.com/support/privacy. Only when Android or iOS selects Firebase may Google LLC process device and system information, App version, network/IP, a fixed whitelist of events, and processed crash diagnostic information; Firebase Performance, advertising personalization, and advertising attribution are currently disabled.
- Model files are distributed on demand by third-party model hosting services such as ModelScope, Hugging Face, or GitHub. This use only calls their file download, resumable transfer, and connection security capabilities, and the relevant services may receive the standard network metadata required to complete these operations; they do not receive recordings, transcripts, meeting titles, search content, speaker information, or local file paths, and no third-party AI inference interfaces are called. Depending on region, model, and dependencies, actual requests may be sent to only some of these services; the App will separately explain and request consent before starting a download. Through minimized transmission, purpose limitation, and technical configuration, we limit the scope of contact with model hosting services to the network metadata described above and require the relevant third parties to protect it at a standard no lower than this Policy and applicable law.
- Whether Umeng or Firebase is selected, the relevant services are used only for limited usage analytics, crash reporting, and basic performance analysis. We do not actively send recordings, imported audio, transcripts, speaker information, meeting titles, search terms, feedback content, tokens, request bodies, installation UUIDs, internal device IDs, task IDs, recording IDs, transcript IDs, raw routes, or dynamic route parameters.
- Apple App Store and StoreKit handle products, payments, subscriptions, restoring purchases, and refunds; Scribo receives only the information required to verify transactions and grant entitlements, and Apple processes payment information under its own rules.
- The Scribo API runs on Alibaba Cloud cloud servers, network, and database environments, and is used to process device sessions, entitlements, transactions, feedback, and necessary operational logs; it does not host a recording or transcript content library.
The actual fields collected by third-party SDKs depend on platform, version, and configuration. We will review permissions, privacy manifests, and platform privacy labels against the production release package; the design or historical implementation of later features does not mean that the corresponding SDK is included in the current release package.
4. Retention, Deletion, and Cross-Border Processing
- Local recordings, imported audio, transcripts, and derived content are generally retained in the current installation space until you actively delete them, uninstall the App, or clear App data, or until the device system processes them under its rules. System backups, exported files, and shared copies may not be controlled by Scribo.
- Device tokens are currently valid for thirty days. Device sessions, entitlements, transactions, feedback, and security logs are retained for the period necessary to provide the Service, perform reconciliation, prevent fraud, handle disputes, and meet legal requirements; when the processing purpose has been achieved and no law requires continued retention, we will delete or anonymize the information.
- You can delete local records in the App and may also request processing of personal information controlled by us through the contact email. Information that must be retained for legal, transaction verification, security audit, or dispute-handling reasons may not be deleted immediately.
- ModelScope generally provides model files in mainland China; when accessing Hugging Face or GitHub, or when Firebase is selected on Android/iOS, IP addresses, network connection information, and analytics or crash metadata may be processed by overseas service providers. We currently do not send recordings, imported audio, transcripts, or other core meeting content to overseas service providers.
5. Security and Your Rights
- We adopt measures such as HTTPS, app sandboxing, system secure storage, access control, transaction signature verification, field whitelists, and log redaction according to the processing scenario, but no security measure can guarantee absolute security.
- You can revoke system permissions such as the microphone, delete local content, and contact us to exercise your lawful rights to access, copy, correct, delete, withdraw consent, restrict processing, request explanations, and complain. Withdrawing consent does not affect processing already carried out based on consent before the withdrawal. This Policy does not exclude or limit any mandatory personal-information-protection rights available to you under the laws of your habitual residence.
- To verify requests and protect information security, we may ask for necessary information related to the request; we will not ask you for payment passwords, complete recordings, or complete transcripts. We will process requests within the period required by applicable law.
- The Service is not primarily directed at children under 14 years of age. Processing children’s personal information requires consent lawfully given by a guardian; guardians may request verification or processing of the relevant information through the contact email.
6. Policy Updates and Contact
The personal information processor is the Scribo team. If the processing purposes, information types, third-party services, cross-border situations, or operational arrangements change materially, we will update this Policy and, where required by law, prompt you again or obtain consent through the App. For privacy questions, complaints, or personal information rights requests, contact us through “Contact Us” in the App or by email at yzx_ios@163.com.