Security Policy
Purpose
This policy sets the security requirements for AI systems at Rygen Technologies, to protect sensitive data throughout the AI lifecycle and keep AI deployments secure.
Scope
This policy applies to all AI systems and tools used at Rygen:
- X1 Platform AI features
- Corsair TMS AI capabilities
- Internal AI tools and third-party AI services
Security Framework
We apply defense in depth to AI systems, securing every layer from data handling through model deployment and output validation.
Data Protection
- Input confidential, proprietary, or client data only into external AI tools and services approved for that data under AI-016 and AI-020, and use them within the restrictions set at approval
- AI models and services used inside an in-scope AI system are approved for that system’s data through AI-013 (Decision Authority), with the provider and its terms recorded in the system’s design specification; they are not listed in AI-020.1
- Treat all AI tool interactions as potentially public
- Follow existing SOC2 data classification and handling procedures for AI systems
- Use only the minimum data an AI system needs to function
Access Controls
- Use only AI tools approved under AI-020
- AI system access restricted to authorized personnel
- Third-party AI APIs accessed through secure, authenticated connections
System Security
- AI models and data stored in secure, monitored environments
- Production AI deployments are approved by the authorities AI-013 Decision Authority names
- Vulnerability scanning and security monitoring applied to AI infrastructure
Output Validation
- Review and validate output from AI tools before business use; for AI systems Rygen builds, review and validate outputs as each system’s design specification and risk assessment require
- Monitor AI systems for failures through GCP monitoring and provider status-page alerts
- Maintain human oversight for all critical AI-driven decisions
Responsibilities
- Principal AI Engineer: owns the policy and oversees it
- Information Security Officer: approves new AI tools, maintains the approved AI tools list (AI-020.1) and security monitoring
- All staff: comply with these security requirements
Violations
Violations may result in disciplinary action including training requirements, suspension of AI access privileges, or termination, depending on severity.
Revision History
| Version | Date | Author | Summary of Change |
|---|---|---|---|
| 1.0 | 2025-06-05 | Field Bradley | Initial draft. |
| 1.1 | 2025-09-02 | Field Bradley | Migrated to markdown and gitlab |
| 1.2 | 2026-11-05 | Field Bradley | Data protection allows confidential, proprietary, or client data only in external AI tools and services approved for it under AI-016 and AI-020, matching AI-016 (AI-1976) |
| 1.3 | 2026-11-09 | Field Bradley | Shortened the policy, with tool approval under AI-020, system output review as each design specification sets, and production deployment approved under AI-013 (AI-2173, AI-2131). States how AI systems are monitored for failures. |