aviator-agents-data-usage-and-retention-policy.md

Aviator Agents Data Usage & Retention Policy

Models Used

We primarily use Claude Sonnet 3.7 for LLM-based tasks internally. While other models may be evaluated in the future, our primary dependency is on Claude Sonnet 3.7.

Data Retention Policy

• We do not save the code generated during a migration.
• We do retain chatbot conversation history for 1 year. Not applicable for on-prem.
• No user data or generated code is used for retraining or fine-tuning models.
• Logs are maintained for compliance and monitoring but we do not store code snippets. Logs are retained for 30 days.

Data Usage & Security Guidelines

1. No Training on User Data
   • We ensure that our LLM does not use user-provided inputs for training.
   • Conversations and code shared are not incorporated into model updates or fine-tuning.
   • More info: Anthropic’s API Data Retention Policy
2. Copyrighted Code Assurance
   • The latest version of the Claude API offers a copyright assurance similar to that of GitHub Copilot
     • Customers have ownership rights to generated output and legal protection against copyright claims to AI generated code.
     • More info: Anthropic’s Copyrighted Code Assurance Policy