Privacy Policy

1. Introduction

Welcome to Talkos (“we,” “us,” or “the App”). We understand that AI language learning involves highly personal data—especially your voice and conversation content. Privacy protection is therefore a core principle in our product design.

This Privacy Policy explains:

• What information we collect and why
• How AI processes and uses your voice data
• The high level of control you have over your data
• How we protect the privacy of your conversations

By using the App, you agree to the terms of this Privacy Policy.

2. Information We Collect

2.1 Information You Provide

Learning and Profile Information

• Voice Input: Audio recordings generated when you interact with an AI tutor or practice pronunciation.
• Learning Preferences: Your native language, target language, proficiency level, and topics of interest (e.g., travel, daily conversation).
• Conversation Context: Text inputs or scripts you provide to guide role‑play scenarios.

Payment Information

• Subscription Selection: We do not store your credit card or payment credentials. Payments are processed independently by third‑party providers such as Apple or Google.

2.2 Information Collected Automatically

Account Information

• Anonymous Device ID: Automatically generated to identify your account and sync learning progress.
• Minimal Identity Data: Depending on your login method, we may store encrypted authentication tokens. Core features do not require your real name or phone number.

Device Information

• Device model and operating system version (to ensure audio compatibility)
• Microphone specifications (such as sample rate and channels, to optimize speech recognition)
• Screen resolution (to optimize interface layout)

Usage Data

• Feature usage (e.g., frequently used lessons or scenarios)
• Performance data (crash logs, response latency)
• Learning progress metrics (vocabulary mastery, speaking fluency scores, check‑in records)

2.3 Information We Do Not Collect

• Background Audio: We only record audio when you actively tap the microphone or enter a call mode.
• Voiceprints or Biometric Identifiers: We do not create or store voice fingerprints for identity verification.
• Contacts, SMS, or Call Logs
• GPS or Precise Location Data

3. How We Use Information

3.1 AI Conversations and Feedback Generation

Voice Processing Workflow

• Input: Audio is captured only when you explicitly initiate speaking.
• Analysis: Speech‑to‑Text (ASR), Pronunciation and phoneme analysis, Semantic and intent understanding.
• Generation: The AI generates contextual responses and converts them into audio using Text‑to‑Speech (TTS).
• Isolation: Processing occurs in an encrypted environment. Audio is deleted after processing or securely stored based on your settings.

Data Isolation Commitment

• Your conversation data is logically isolated from other users’ data.
• By default, our staff cannot access or listen to your audio recordings, unless you explicitly submit them as part of a support request or disclosure is required by law.

3.2 Service Improvement

• We use de‑identified, aggregated statistics (e.g., common pronunciation difficulties) to improve courses.
• Crash logs and performance metrics are used to fix technical issues.

3.3 Notifications and Communications

• Practice reminders
• Periodic progress summaries

You may disable non‑essential notifications in the app settings at any time.

3.4 Security and Abuse Prevention

• Automated mechanisms are used to detect bots or clear abuse patterns.
• These mechanisms are used solely to protect service integrity and are not used for advertising, marketing, or user profiling.

4. Special Notes on AI Models and Data Use

4.1 AI Training Data

We commit that:

• Base models are trained on publicly available or properly licensed datasets.
• By default, your personal voice recordings are not used to train general AI or speech recognition models.
• If we ever seek to use user audio to improve specific capabilities (e.g., regional accent recognition), we will request your explicit opt‑in consent.
• Your data will not be used for advertising, user profiling, or commercial targeting.

4.2 Rights to Generated Content

• You retain full ownership of your own voice.
• You have full usage rights to AI‑generated corrections, scripts, and feedback provided for learning purposes.

4.3 Voice Analysis Technology

We analyze linguistic content (grammar, pronunciation, semantics) only. We do not infer your identity, biometric traits, age, or gender from your voice, unless you explicitly enable a learning feature that requires such information.

5. Information Sharing and Disclosure

5.1 We Do Not Sell Your Data

We do not sell, rent, or trade your personal information, voice recordings, or conversation content.

5.2 Essential Service Providers

We share the minimum necessary data with essential providers, including:

• Cloud infrastructure and AI inference services
• Crash reporting and system reliability services

All data is transmitted using encryption and protected by contractual obligations prohibiting model training or secondary use.

6. Data Storage and Security

6.1 Retention Periods

• Active Accounts: Necessary profile and learning data are retained while your account remains active.
• Voice Audio: Raw audio is not stored long‑term by default. Only transcripts and performance metrics are retained.
• Optional Storage: If you choose to save a lesson, the specific audio file is encrypted and stored until you delete it.

6.2 Deletion

• After account deletion, personal data is removed from active systems within 30 days.
• Backup data is deleted within 90 days.

7. Your Rights and Controls

You have a high degree of control over your personal data, including the right to:

• Access your learning history and records
• Delete individual conversations or your entire account
• Withdraw microphone access at any time (which may limit speaking features)

Limitations for Anonymous Users:

If you use the App anonymously (without logging in via email or third-party auth), we may not be able to verify your identity to fulfill specific data access or deletion requests (e.g., "Delete all my data"). In such cases, the most effective way to exercise your right to deletion is to delete the App from your device, which will sever the link between your device and the anonymous data stored on our servers. Inactive anonymous data is automatically purged after 90 days.

8. Children’s Privacy

• Talkos is intended for general audiences and is rated 12+ on app stores. We do not knowingly collect personal information from children under 13. If you are under 13 years old, please do not use the App or provide any information without the supervision and verifiable consent of a parent or legal guardian.
• Since Talkos allows anonymous usage without age verification, parents and guardians are solely responsible for monitoring their children's use of the App. We treat all anonymous interaction data as standard user data. If a parent discovers their child has provided personal information (such as voice recordings containing personal details) via an anonymous account, please contact us with the relevant Device ID (found in Settings) so we can locate and delete the data.
• If we become aware that we have collected personal information from a child under 13 without verification of parental consent, we will take steps to remove that information from our servers immediately.

9. Updates to This Privacy Policy

• Material Changes: We will provide at least 30 days’ notice via in‑app notification.
• Minor Updates: Clarifications will be reflected in the “Last Updated” date.

You may choose to accept the updated policy or discontinue use of the App.

10. Applicable Law and Cross‑Border Data Transfers

This Privacy Policy is intended to comply with applicable data protection laws, including but not limited to the GDPR and CCPA/CPRA.

Because we use global cloud infrastructure, your data may be processed outside your country or region. We implement appropriate safeguards required by applicable law.

11. Our Commitment

• We do not record audio in the background
• We do not create voiceprints or deepfake models
• Human access to your audio practice is restricted by default
• Deleted content is removed within a reasonable timeframe
• Your voice is used for learning—not surveillance or commercialization