Data Usage Policy

How we use your data to power Clare's AI and improve your recovery support

Last updated: January 2025

1. Overview

Clare's effectiveness as an AI Recovery Coach depends on understanding emotional context, recovery patterns, and personalized support needs. This Data Usage Policy explains how we responsibly use your information to:

  • Power Clare's emotional intelligence and conversation capabilities
  • Detect crisis situations and provide appropriate support
  • Personalize your recovery experience
  • Improve our AI models to help future users
  • Ensure platform safety and effectiveness

All data usage follows strict privacy protections and ethical AI principles.

2. Real-Time AI Processing

How Clare Understands You

Every message you send to Clare is processed in real-time to understand emotional context, detect potential concerns, and generate appropriate, empathetic responses.

Text Conversation Processing:

  • Emotional Analysis: Messages are analyzed for emotional tone, distress levels, hope indicators, and confidence measures
  • Crisis Detection: Advanced algorithms scan for suicidal ideation, severe depression, or relapse risk indicators
  • Context Understanding: Previous conversations inform current responses for continuity and personalization
  • Response Generation: AI models create empathetic, recovery-focused responses tailored to your current emotional state
  • Safety Monitoring: Ongoing assessment for concerning patterns or escalating issues

What This Means:

  • Your messages are processed immediately to provide relevant support
  • Emotional patterns help Clare understand your unique recovery journey
  • Crisis detection can trigger immediate resource provision
  • Conversation history enables more meaningful, connected support
3. Voice Data Processing (Guardian Tier)

Voice Calling Technology

Voice calls with Clare use HumeAI's emotional intelligence platform to analyze vocal patterns, tone, and emotional indicators for deeper understanding and support.

Voice Processing Pipeline:

1. Real-Time Analysis
  • • Voice tone and inflection analysis
  • • Emotional state detection
  • • Stress level assessment
  • • Speech pattern recognition
2. Transcription
  • • Speech-to-text conversion
  • • Content understanding
  • • Keyword and phrase analysis
  • • Integration with text processing
3. Response Generation
  • • Emotionally-informed AI responses
  • • Voice synthesis with appropriate tone
  • • Crisis escalation if needed
  • • Personalized recovery support
4. Data Handling
  • • Conversation summary storage
  • • Audio deletion after processing
  • • Emotional insights retention
  • • Privacy protection measures
4. AI Model Training and Improvement

Your interactions help make Clare smarter and more effective for all users in recovery. Here's how we responsibly use data for AI improvement:

What We Use

  • • Anonymized conversation patterns
  • • Successful intervention strategies
  • • Emotional response effectiveness
  • • Crisis detection accuracy
  • • Recovery outcome correlations

What We DON'T Use

  • • Personal identifying information
  • • Raw voice recordings
  • • Individual conversation content
  • • Specific personal details
  • • Data from users who opt out

Anonymization Process:

  1. De-identification: All personal identifiers are removed
  2. Aggregation: Individual data points are combined into patterns
  3. Statistical Analysis: Focus on trends rather than individual cases
  4. Privacy Review: Regular audits ensure no re-identification is possible

How This Helps Everyone

By understanding what support strategies work best, we can improve Clare's ability to help people in different stages of recovery, facing various challenges, and with diverse backgrounds.

5. Crisis Detection and Safety

Your safety is our highest priority. We use advanced AI to detect potential crisis situations and provide immediate support:

Detection

AI monitors for suicide risk, severe depression, or relapse indicators

Response

Immediate supportive messaging and resource provision

Escalation

Connection to crisis hotlines and emergency resources when needed

Crisis Data Usage:

  • Pattern Recognition: Learning to identify early warning signs
  • Response Effectiveness: Measuring which interventions help most
  • False Positive Reduction: Improving accuracy to avoid unnecessary alerts
  • Resource Optimization: Ensuring the most helpful resources are provided
6. Personalization and Learning

Clare learns about your unique recovery journey to provide increasingly personalized support:

Personal Learning:

  • Your communication style and preferences
  • Topics that resonate with you
  • Times when you need extra support
  • Recovery strategies that work for you
  • Triggers and warning signs specific to you

Adaptive Responses:

  • Conversation tone adjusted to your needs
  • Recovery advice tailored to your situation
  • Crisis intervention calibrated to your patterns
  • Resource recommendations based on your preferences
  • Support timing optimized for your schedule

Privacy in Personalization

Personal learning data is encrypted and siloed to your account. It's not shared with other users or used in general AI training without anonymization.

7. Your Control and Choices

You have control over how your data is used:

✅ You Can:

  • View all data we have about you
  • Delete your conversation history
  • Opt out of AI training contributions
  • Adjust personalization settings
  • Export your data
  • Delete your account entirely

⚙️ Settings Available:

  • Crisis detection sensitivity
  • Data retention preferences
  • AI training participation
  • Personalization level
  • Voice processing options
  • Data sharing permissions

Easy Controls

Access all privacy controls through your account settings. Changes take effect immediately and you can modify them anytime.

8. Research and Ethical AI

We conduct research to improve recovery support technology while maintaining the highest ethical standards:

Research Goals

  • • Improving crisis detection
  • • Understanding recovery patterns
  • • Developing better support strategies
  • • Reducing relapse rates

Ethical Principles

  • • Privacy by design
  • • Informed consent
  • • Transparency in AI decisions
  • • Bias detection and mitigation

Safeguards

  • • IRB review for research
  • • Regular ethics audits
  • • User opt-out options
  • • Data minimization
9. Questions and Updates

We update this policy as our AI capabilities evolve. We'll notify you of significant changes and always maintain transparency about data usage.

Contact Us About Data Usage:

  • Email: data@claritywellness.app
  • Privacy Officer: privacy@claritywellness.app

Request Data Actions:

  • Account settings for immediate changes
  • Email support for complex requests