YiTu Privacy Policy

Plain-language summary

• What we process: When you use cloud voice features, we process speech from you and people nearby, plus English or Chinese transcripts, translations, and returned speech created from that audio. We also process limited technical data needed to connect and troubleshoot. The App stores two integers on the device: the adult-eligibility version and cloud-processing consent version.
• Who receives it: Voice features use YiTu cloud services hosted by Cloudflare and third-party AI provider OpenAI. Google processes your email address, message, and attachments only when you choose to send a support or privacy email. The static privacy site is hosted on Cloudflare Pages; Web Analytics is not enabled, no Pages Functions are used, and the operator has not configured site analytics or application-level request logging.
• Why: Only to recognize speech, translate, generate and play speech, connect and secure the service, troubleshoot, answer email, and meet legal obligations. AI is not used to profile you or make decisions with legal or similarly significant effects.
• Default retention: Captions, history, and returned audio remain in App runtime memory. Temporary “Speak for me” M4A files are deleted on a best-effort basis. Adult eligibility and consent versions remain on the device. The YiTu Worker has no business-content database or object storage. OpenAI defaults are in Section 6; some default abuse-monitoring logs may be kept for up to 30 days. Email is kept only as needed to handle the request, protect rights, or meet legal obligations.
• Main risks: Speech may contain sensitive information about you or others. Network transmission and third-party processing cannot be risk-free. Data may be processed in the United States and other locations published by providers, where courts, law enforcement, or national-security authorities may obtain lawful access. Avoid saying information that is unnecessary for translation, and use the service only when you may lawfully provide nearby speech.

This summary highlights the main points. The full policy below controls.

1. Scope, operator, and service region

This policy applies to the YiTu／译途 iOS app (the “App”), YiTu services that support the App’s cloud features, and the static website that publishes this policy.

The operator and Privacy Officer are HAICHUAN ZHANG. The V1 user service region is fixed to the United States and Canada, including Quebec. Mainland China, Hong Kong, Macao, and all other regions are outside V1.

HAICHUAN ZHANG handles operational administration and privacy requests from Canada. The Canadian administration location and the user service region are separate facts; administration from Canada does not mean providers process data only in Canada.

The App has no user accounts. We do not require a name, email address, phone number, date of birth, or precise location to provide translation.

2. Age 18+ and consent to cloud processing

V1 cloud features are only for people aged 18 or older. Before entering a core mode, the App asks you to confirm adult eligibility and then to consent to cloud processing. If you decline or withdraw consent, the relevant cloud features will not start or will stop. The App will not request a new temporary credential, upload a new “Speak for me” recording, or establish a new Realtime session.

The App does not request or store a birth date or exact age. Adult eligibility and cloud consent are separate records. Withdrawing cloud consent does not delete the adult-eligibility record. You can withdraw at any time under “Privacy & Data”; cloud features will ask again the next time you use them.

3. How the four voice modes work

1. Listen only: The App sends English speech directly to OpenAI Realtime, which creates the English transcript and Simplified Chinese translation. It does not generate or play Chinese translated speech.
2. Speak for me: While idle, the App automatically listens to English and displays the English transcript and Simplified Chinese translation without playing Chinese translated speech. When you press and hold to speak Chinese, the App first stops English listening and sends the Chinese recording to the YiTu Worker hosted on Cloudflare. The Worker forwards it to OpenAI to create, in sequence, a Chinese transcript, English translation, and English speech, then returns the text and audio to the App for playback. Idle English listening may resume after playback.
3. Listen with headphones: The App sends speech to OpenAI Realtime, receives transcripts and translations, and plays translated speech through connected headphones.
4. Two-way conversation: The App uses OpenAI Realtime for live Chinese and English speech, generates a source-language transcript and translation into the other language, and plays translated speech.

Nearby speech may belong to another person. Use the App only when you are entitled to provide that speech, and give notice or obtain permission when required by the circumstances and applicable law.

4. Data we process

4.1 Speech and speech-derived content

• English or Chinese speech captured by the microphone;
• English or Chinese transcripts, captions, and translations;
• English or Chinese speech generated and returned by OpenAI;
• names, health, financial, address, or other information that you voluntarily include in free-form speech. YiTu does not ask you to provide it.

4.2 Feature, connection, and diagnostic data

To connect, protect, and troubleshoot the service, the App, YiTu Worker, and providers may process the feature mode, language direction, temporary session or request identifiers, times and processing stages, network status, IP address, request headers, error category, status code, audio or text length, returned-data size, and device-side states such as microphone permission, network type, and audio route.

App diagnostic logs do not record the contents of speech, captions, transcripts, translations, or returned audio, and do not record access-credential contents. Worker application logs do not intentionally record this business content.

4.3 On-device records

The App separately stores in iOS UserDefaults:

• an adult-eligibility version integer; and
• a cloud-processing consent version integer.

These records do not contain a name, birth date, exact age, confirmation time, account identifier, or advertising identifier.

4.4 Static-website access data

The static pages do not set cookies, run Web Analytics or other site analytics, load third-party fonts, scripts, images, or other third-party runtime resources, or offer forms. The Pages site publishes only static files and uses no Pages Functions or application Worker; the operator has not configured site analytics or application-level request logging and does not collect or retain access logs through site code or Pages Functions. Cloudflare still processes platform network metadata to deliver pages, maintain security, and troubleshoot, such as IP addresses, request URLs and methods, times, routing and system-configuration information, request headers, response status, and error information.

The static privacy site is hosted on Cloudflare Pages. Cloudflare’s official documentation states that Direct Upload deploys static assets to its global network, so processing must not be described as occurring only in Canada or any single location. Cloudflare may process and retain Customer Logs, Network Data, and other platform records under the applicable contract, its Privacy Policy, security needs, and legal obligations. The available official documentation does not state one exact retention period that applies to every platform record for this static Pages site, and the operator does not invent one.

Cloudflare’s edge may also send NEL and Report-To response headers. When a qualifying network failure occurs, the browser may report network-failure information to Cloudflare, including the request URL, method, referrer, protocol, status, and error type. Cloudflare’s official NEL documentation states that it derives ASN, country, and metro area from the reporting request’s IP and drops the client IP from the NEL pipeline after the request is processed; this does not mean Cloudflare processes no other platform network metadata.

4.5 Email you choose to send

If you email privacy@yituapp.com for support, privacy, or security, we process your email address, message body, email metadata, and attachments you choose to add. This formal mailbox is provided through Google Workspace and supports external receiving, sending, and direct replies in Chinese, English, and French. Google, as the email provider, processes this data to send, receive, and store mail, filter spam or malicious content, and protect the account.

We use email data only to answer and manage the request, verify its scope, maintain necessary records, protect rights and security, and meet legal obligations. Do not send identity documents, audio, or other sensitive attachments unless they are necessary for the request.

5. Purposes and limits on AI use

We process data only to:

1. provide speech recognition, transcription, captions, translation, speech generation, and playback;
2. establish, maintain, end, and protect Realtime or upload connections;
3. apply the selected mode, language direction, adult eligibility, and consent settings;
4. detect, troubleshoot, and fix errors, prevent abuse, and maintain security;
5. answer support, privacy, or security email and handle rights requests; and
6. comply with applicable law, court orders, or valid legal process.

Where applicable law requires it, cloud processing of speech and speech-derived content is based on your consent. Providing the translation you request is also the purpose of processing. Security, troubleshooting, and legal records are limited to what is needed to maintain the service and meet obligations.

AI is used only for speech recognition, translation, and speech generation. YiTu does not use this data to profile users, assess eligibility for credit, employment, housing, insurance, education, or healthcare, or make decisions with legal or similarly significant effects.

6. Recipients, subprocessors, and default retention boundaries

We disclose data only to recipients needed to provide the service. OpenAI, Cloudflare (including the Worker used by voice features and Pages used by the static privacy site), and Google may use affiliates and necessary subprocessors to host, transmit, protect, and support their services. Those subprocessors should receive only data needed for the relevant service.

YiTu selects only providers that give user data the same or equivalent protection required by this policy and applicable law. Under service terms, data processing addenda, or other written arrangements that actually apply to the relevant account and service, providers must implement corresponding data-protection measures. Provider lists and locations may change under their published terms. Using a provider does not relieve HAICHUAN ZHANG of responsibility for protecting user data or for provider selection, instructions, and ongoing oversight. We do not claim to have a separate customized contract that does not actually apply.

6.1 YiTu App and Cloudflare Worker

• In Realtime modes, media generally travels directly from the App to OpenAI using a short-lived credential; the Worker obtains that credential.
• Chinese recordings, Chinese transcripts, English translations, and English speech for “Speak for me” pass through the Worker.
• The Worker processes business content in memory during a request, has no configured business-content database or object storage for speech, transcripts, translations, or returned audio, and marks business responses not to be cached.
• Worker application logs contain limited metadata such as request identifier, path, step, status, duration, size or length, and error code; they do not intentionally include business content or credential contents.

Cloudflare supplies network, runtime, security, and logging infrastructure for the Worker used by the App’s voice features and may therefore process content passing through that Worker and connection metadata. The operator has not enabled Workers Logs or Traces for that Worker. Under Cloudflare’s current Workers Logs documentation, if Workers Logs is enabled for the Worker in the future, default retention is 3 days on Workers Free and 7 days on Workers Paid, with a platform maximum of 7 days. This is a Worker logging rule and does not apply to the static Pages requests described below. Other Cloudflare platform records have no single fixed period proven by this code and follow account settings, Cloudflare’s published terms, legal requirements, and deletion processes.

See Cloudflare Workers Logs, Cloudflare Privacy Policy, and Cloudflare service subprocessors.

6.2 OpenAI

Depending on the mode, OpenAI receives speech, speech-derived text, session instructions, language settings, and network and session metadata needed to connect, and returns transcripts, translations, and/or speech. YiTu data flows use Realtime, audio transcription, Responses, and speech-generation endpoints.

According to OpenAI’s current official data-controls documentation, API data is not used to train models by default unless the account opts in to sharing. The organization used by YiTu has feedback, evaluation/fine-tuning, and API input/output sharing disabled; the project uses Global residency; and API-call logging is enabled per call, with Responses store:false disabling the corresponding account-visible call log. YiTu does not claim that Zero Data Retention (ZDR), Modified Abuse Monitoring (MAM), or regional data residency is enabled.

store:false is not ZDR and does not disable default abuse monitoring. OpenAI may retain relevant data longer when legally required or reasonably necessary to prevent severe harm. See OpenAI data controls.

6.3 Google/Gmail

Google/Gmail processes the Section 4.5 data only when you choose to email us. Email remains in the operator’s Gmail mailbox until the request is complete and it is no longer needed for follow-up, rights protection, security, or legal obligations; it is then deleted or de-identified. You may also request deletion, subject to legal-retention and security exceptions. Google explains that deletion may take time to propagate between active and backup systems and that some data may be kept for security, anti-fraud, or legal purposes. See the Google Privacy Policy.

6.4 Cloudflare Pages static-site hosting

The static privacy site is hosted on Cloudflare Pages and publishes only static files; it does not use Pages Functions. Web Analytics is not enabled, and the operator has configured no site analytics or application-level request logging. Pages Functions metrics and real-time logs apply to Function invocations; this static site has no such application invocation logs, and the 3/7-day Worker logging rule above does not apply to static Pages requests. Cloudflare may still process platform network metadata needed for delivery, security, errors, and NEL in its global infrastructure under its contracts and published policies. Official documentation does not establish one exact retention period or one processing location for every such platform record.

7. App retention and deletion

• Captions, transcripts, translation history, and returned speech remain only in runtime memory. They are not written to a persistent App business-history database and do not remain there after the relevant page or lifecycle ends or the App process terminates.
• “Speak for me” creates an M4A file in the system temporary directory. The App attempts deletion after a short recording, request completion or failure, cancellation, leaving the page, entering the background, consent withdrawal, or destruction of the relevant object. Deletion can fail, and a crash or forced termination can interrupt cleanup, so this is best-effort deletion rather than an absolute immediate-deletion guarantee.
• Returned “Speak for me” audio is played from memory and is not saved as a persistent App audio file.
• The adult-eligibility version remains until App data is cleared. The cloud-consent version remains until you withdraw, App data is cleared, or the required consent version changes.

8. Cookies, analytics, ads, sale, sharing, and DNT

The YiTu App and static website contain no advertising SDK, cross-app or cross-site tracking, third-party analytics, or data-broker feature. We do not sell personal information or “share” it for cross-context behavioral advertising.

Because the App and website do not conduct cross-site or cross-app tracking, a browser Do Not Track (DNT) signal does not change our existing practices. The website does not automatically choose a language based on location or silently redirect you. You can directly open or manually switch among Chinese, English, and French pages.

9. Cross-border processing and government-access risk

Data may be processed in the United States and other processing locations published by OpenAI, Cloudflare (including Pages), Google, and their subprocessors. In particular, Realtime speech and Worker processing may occur outside your state, province, or country. The operator handles administration and privacy requests from Canada, but this does not limit provider processing to Canada.

Privacy laws may differ abroad. Foreign courts, law-enforcement agencies, or national-security authorities may obtain lawful access under local law. We use applicable processor selection, purpose and data limits, contracts, and other available safeguards, and respond to questions about cross-border processing. Using processors does not relieve HAICHUAN ZHANG of responsibility for assessing their safeguards.

10. Your choices and rights

You may:

• decline or withdraw cloud-processing consent under “Privacy & Data”;
• decline or withdraw microphone permission in iOS settings;
• email privacy@yituapp.com to request access to, correction of, or deletion of personal information we hold;
• withdraw applicable consent and object to or restrict applicable processing;
• where applicable law provides a data portability right, obtain a portable copy;
• complain about our processing; and
• where applicable U.S. state law provides, appeal a decision on a rights request and exercise rights without discrimination.

The App has no account, and session content in runtime memory disappears quickly. We generally cannot locate an anonymous voice session using only a name or email address. To help investigate, you may voluntarily provide an approximate date, time, mode, and error information. Do not send the contents of speech, captions, or translations unless necessary. We do not create an additional user profile to handle a rights request or ask for information beyond reasonable verification.

If data is held only by a provider within default logging boundaries, we will explain our ability to locate and delete it and, where applicable, pass the request to the processor. We cannot promise individual deletion of data the provider cannot reasonably identify.

11. Canadian complaint channels

Please first contact the Privacy Officer at privacy@yituapp.com. You may write in Chinese, English, or French, and we will handle the request in a language reasonably suited to it. Our public Privacy Governance and Request Handling page explains responsibility, retention/destruction, and complaints. If the matter is not resolved, users in Canada may seek information or complain to:

• Office of the Privacy Commissioner of Canada (OPC): official concern and complaint portal.
• Commission d’accès à l’information du Québec (CAI): official remedies and complaint portal.

You may also complain to another competent authority where you live. Acceptance depends on applicable law and the authority’s jurisdiction.

12. Security

We use limited measures supported by the current implementation: HTTPS, short-lived Realtime credentials, encrypted Realtime media negotiation, not returning the long-lived server key to the App, data minimization by feature, excluding business content and credential contents from application logs, and stopping microphones and cloud sessions when their lifecycle ends.

No transmission, processing, or deletion method is completely secure. Do not include sensitive information in speech or email unless needed for the service. Contact privacy@yituapp.com if you suspect a privacy or security incident.

13. Policy changes

We may update this policy when features, providers, law, or operations change and will update the version and dates at the top. If purposes, recipients, or risks materially change, we will obtain consent again through the App where appropriate.

14. Contact

• Product: YiTu／译途
• Operator: HAICHUAN ZHANG
• Privacy Officer: HAICHUAN ZHANG
• Email: privacy@yituapp.com
• Website: https://www.yituapp.com/
• Privacy policy: https://www.yituapp.com/privacy/
• Privacy governance and request handling: https://www.yituapp.com/en/governance/
• Operational administration location: Canada

The Chinese, English, and French versions are intended to provide materially consistent information. If you find an inconsistency, contact the Privacy Officer. We will correct it using the interpretation that best protects your rights under applicable law and reflects actual processing.