Problem
SafeVoice allows users to anonymously share experiences related to harassment, abuse, emotional distress, and other sensitive situations.
Although the repository already contains an AI-based crisis classification service, it is not integrated into the story submission workflow. As a result, potentially high-risk content can be published without identifying users who may need immediate support resources.
This creates a gap between the platform's mission and its current behavior.
Proposed Solution
Integrate the existing crisis classification pipeline into the story creation flow.
Workflow
User submits story
↓
AI Crisis Classification
↓
Assign Risk Level
↓
Store Classification Metadata
↓
Display Support Resources (if needed)
Requirements
Risk Classification
Classify submitted stories into:
using the existing AI classification endpoint.
Metadata Storage
Store classification information alongside stories:
risk_level
risk_reason
classified_at
Emergency Support Resources
For HIGH and CRITICAL risk stories:
- Display contextual support resources
- Surface relevant helplines and NGO resources
- Encourage users to seek immediate assistance
Graceful Fallback
If classification fails:
- Allow story submission
- Use a safe default value
- Log the failure
- Do not block the user workflow
Expected Impact
- Improves user safety and platform responsibility
- Activates existing AI infrastructure already present in the project
- Provides timely support to users experiencing crises
- Enhances the overall trust and usefulness of the platform
Files Likely Affected
src/pages/ShareStory.tsx
netlify/functions/classify-crisis.cjs
src/pages/Home.tsx
src/pages/Stories.tsx
Problem
SafeVoice allows users to anonymously share experiences related to harassment, abuse, emotional distress, and other sensitive situations.
Although the repository already contains an AI-based crisis classification service, it is not integrated into the story submission workflow. As a result, potentially high-risk content can be published without identifying users who may need immediate support resources.
This creates a gap between the platform's mission and its current behavior.
Proposed Solution
Integrate the existing crisis classification pipeline into the story creation flow.
Workflow
Requirements
Risk Classification
Classify submitted stories into:
using the existing AI classification endpoint.
Metadata Storage
Store classification information alongside stories:
Emergency Support Resources
For HIGH and CRITICAL risk stories:
Graceful Fallback
If classification fails:
Expected Impact
Files Likely Affected