NeuroPro takes security seriously. This document outlines our security practices, how to report vulnerabilities, and our commitment to protecting user privacy.
- 100% On-Device Processing: All AI computations run locally on the user's device
- Zero Cloud Dependency: No data is ever transmitted to external servers
- No Telemetry: No analytics, tracking, or usage data collection without explicit user consent
- No Third-Party SDKs: No advertising, analytics, or tracking frameworks
- No Data Persistence: Camera frames are processed in real-time and immediately discarded
- No Image Storage: No photos or videos are saved without explicit user action
- Secure Memory: Metal buffers are properly deallocated after processing
- No Logs: No sensitive information is logged to console or files
- Type Safety: 100% Swift type-safe code, no force unwraps in production
- Memory Safety: Automatic Reference Counting (ARC) prevents memory leaks
- Input Validation: All user inputs and camera frames are validated
- Secure APIs: Only Apple-provided security-audited frameworks
We release security updates for the following versions:
| Version | Supported |
|---|---|
| 1.0.x | ✅ |
| < 1.0 | ❌ |
- Permission-Based: Requires explicit user permission (iOS standard)
- Indicator Display: iOS automatically shows camera usage indicator
- Revocable Access: Users can revoke permission in iOS Settings at any time
- Sandboxed Execution: Metal shaders run in isolated GPU context
- Validated Pipelines: All compute pipelines are validated at compile time
- Memory Isolation: GPU memory is isolated from system memory
- No Network Access: App does not request network permissions
- Offline-First: Fully functional without internet connection
- No External Requests: No HTTP/HTTPS requests to any servers
- iOS allows users to take screenshots of any app
- Detection results visible in screenshots could contain sensitive information
- Mitigation: Users should be aware screenshots may capture detected objects
- VoiceOver and other accessibility services can read detected object labels
- Mitigation: This is intentional for accessibility, users control these settings
- When attached to Xcode debugger, internal state is visible
- Mitigation: Only occurs during development, not in production builds
We take all security reports seriously. If you discover a security vulnerability:
DO NOT open a public GitHub issue for security vulnerabilities.
Email: security@neuropro.app (preferred) GitHub: Use GitHub Security Advisories (private disclosure)
- Description: Clear description of the vulnerability
- Impact: Potential security impact and affected users
- Reproduction: Step-by-step instructions to reproduce
- Environment: iOS version, device model, app version
- Proof of Concept: Code or video demonstrating the issue (if applicable)
- 24 hours: Initial acknowledgment of report
- 72 hours: Preliminary assessment and severity rating
- 7 days: Detailed response with fix timeline or mitigation plan
- 30 days: Security patch released (for confirmed vulnerabilities)
Critical (CVSS 9.0-10.0)
- Remote code execution
- Data exfiltration
- Complete device compromise
High (CVSS 7.0-8.9)
- Privilege escalation
- Unauthorized access to camera
- Sensitive information disclosure
Medium (CVSS 4.0-6.9)
- Denial of service
- Information leakage
- App crash with data loss
Low (CVSS 0.1-3.9)
- Minor information disclosure
- UI spoofing
- Performance degradation
- Keep iOS Updated: Always use the latest iOS version
- App Updates: Enable automatic app updates
- Review Permissions: Periodically review app permissions in Settings
- Passcode/Face ID: Use device authentication to prevent unauthorized access
- ❌ Request internet/network access
- ❌ Store photos without your action
- ❌ Send data to servers
- ❌ Track your location
- ❌ Share data with third parties
- ❌ Display advertisements
- ❌ Collect analytics
- ✅ Process camera frames in real-time
- ✅ Use Metal GPU for performance
- ✅ Run AI models on Neural Engine
- ✅ Discard frames after processing
- ✅ Respect your privacy
- Date: January 2025
- Scope: Full codebase review
- Findings: No critical or high severity issues
- Tools Used: Xcode Static Analyzer, SwiftLint, Manual Review
- All pull requests undergo security review
- Automated static analysis on every commit
- Manual code review for sensitive changes
- Regular dependency updates
- GDPR Compliant: No personal data collection
- CCPA Compliant: No data sale or sharing
- COPPA Compliant: Safe for all ages (no data collection)
- ✅ App Store Review Guidelines 2.5.13 (Privacy)
- ✅ Human Interface Guidelines (Permissions)
- ✅ App Privacy Details (No data collection)
- Metal (GPU framework)
- MetalPerformanceShaders (Neural networks)
- Vision (Object detection)
- Core ML (Machine learning)
- AVFoundation (Camera access)
- SwiftUI (User interface)
All frameworks are:
- Provided by Apple
- Regularly security-audited
- Updated via iOS system updates
- No third-party dependencies
- None: Zero third-party libraries or SDKs
Before every release:
- Static analysis passes (Xcode Analyzer)
- No compiler warnings
- No force unwraps in production code
- All network code removed (verify with network link map)
- Camera permission strings up to date
- Privacy manifest accurate
- Memory leaks checked (Instruments)
- Code signing verified
- Entitlements minimized
We support responsible disclosure and will:
- Acknowledge your contribution in release notes (with permission)
- Credit security researchers (unless requested otherwise)
- Coordinate disclosure timeline with reporter
- Not Pursue Legal Action against good-faith security research
Currently, we do not offer a formal bug bounty program. However:
- We deeply appreciate security reports
- We will publicly credit researchers (with permission)
- We may offer recognition in future bounty programs
Security Team: security@neuropro.app General Inquiries: Contact via GitHub issues (non-security only) Privacy Questions: privacy@neuropro.app
We recognize security researchers who help keep NeuroPro safe:
No vulnerabilities reported yet
Last Updated: January 2025 Next Review: Quarterly (April 2025)