What We Log
Transparency is core to our privacy-first approach. Understand exactly what data Fusion AI collects, how long we retain it, and why it's necessary for providing our service.
Privacy-First Logging
We only log what's essential for service operation, security, and compliance. Your actual prompts and AI responses are never stored. All logging can be customized or disabled based on your privacy needs.
Data We Collect
Request Metadata
Debugging, performance monitoring, and security analysis
Data Points
- API endpoint accessed
- Request timestamp
- HTTP method and status codes
- Request/response size
- User agent and IP address
- Authentication method used
Retention Period
Automatically deleted after this period
Usage Analytics
Service optimization, billing, and feature development
Data Points
- Token usage counts
- Model selection frequency
- Provider routing decisions
- Response time metrics
- Feature utilization
- Error rate tracking
Retention Period
Automatically deleted after this period
Security Events
Threat detection, fraud prevention, and compliance auditing
Data Points
- Failed authentication attempts
- Rate limit violations
- Suspicious access patterns
- API key usage anomalies
- Unusual request frequencies
- Security alert triggers
Retention Period
Automatically deleted after this period
Quality Metrics
Service quality assurance and model optimization
Data Points
- Provider response times
- Model accuracy scores
- User satisfaction ratings
- Error classification
- Performance benchmarks
- Routing effectiveness
Retention Period
Automatically deleted after this period
What We Don't Log
Prompt Content
Your actual prompts and questions are not stored in our logs
Response Content
AI-generated responses are not retained after delivery
Personal Data
No personal information from prompts is extracted or stored
File Contents
Uploaded images and documents are processed but not stored
Data Retention Timeline
Real-time
Active request processingData Types
- • Current API requests
- • Active sessions
- • In-flight responses
Purpose
Immediate processing and forwarding
24 hours
Short-term operational dataData Types
- • Recent error logs
- • Performance metrics
- • Cache entries
Purpose
Automated analysis and optimization
30 days
Monthly operational insightsData Types
- • Usage patterns
- • Billing calculations
- • Performance trends
Purpose
Regular reporting and account management
90 days
Quarterly analysis and debuggingData Types
- • Request metadata
- • Error diagnostics
- • Performance history
Purpose
Deep analysis and service improvements
1 year
Annual compliance and auditingData Types
- • Security logs
- • Compliance records
- • Audit trails
Purpose
Regulatory compliance and security reviews
Control Your Logging
Use these API configurations to control what gets logged for your requests:
Standard Logging
Default behavior with all standard logging enabled
# Standard API request with default logging
curl -X POST https://api.mcp4.ai/v1/chat \
-H "Authorization: Bearer sk-fusion-..." \
-H "Content-Type: application/json" \
-d '{
"prompt": "Explain machine learning",
"provider": "neuroswitch"
}'Minimal Logging
Essential logging only for debugging and security
# Request with minimal logging
curl -X POST https://api.mcp4.ai/v1/chat \
-H "Authorization: Bearer sk-fusion-..." \
-H "X-Privacy-Mode: minimal" \
-d '{
"prompt": "Sensitive query",
"provider": "neuroswitch"
}'No Logging
Zero-log endpoint for maximum privacy
# No-logging endpoint
curl -X POST https://api.mcp4.ai/v1/chat/no-log \
-H "Authorization: Bearer sk-fusion-..." \
-H "X-Privacy-Mode: strict" \
-d '{
"prompt": "Confidential query",
"provider": "neuroswitch"
}'