1. Purpose

This policy defines the measures taken to prevent confidential and sensitive data handled in this environment from being used for AI model training or transmitted to unintended third parties.

2. Services in Use

Service	Provider	Usage
Claude Code (CLI)	Anthropic	Code generation, analysis, automation
Gemini API	Google	Text/vision processing via API

3. Data Protection Measures

3.1 Anthropic / Claude Code

(a) Account-Level Settings (claude.ai)

• "Help improve Claude" — OFF
  Chat and coding session data will NOT be used for AI model training.
• "Location metadata" — OFF
  Approximate location data will NOT be used.

(b) Claude Code Telemetry (settings.json)

• CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC = true
  Disables analytics, error reporting (Sentry), and /bug command.

3.2 Google / Gemini API

• Accessed via API key (not consumer product).
• Per Google Cloud API Terms of Service, API input/output data is NOT used for model training by default.
• No additional opt-out action required.

4. Data Retention

Service	Retention	Notes
Claude (Anthropic)	Up to 30 days	Safety review only; not used for training
Gemini API	Per Google Cloud DPA	API data not retained for training

5. Additional Safeguards

• Sensitive credentials are stored locally (not committed to repositories).
• Database files (SQLite) containing confidential data are stored locally and not synced to cloud services.
• Git repositories do not contain API keys or credentials.

6. Exceptions

• Conversations flagged for safety review by Anthropic may be reviewed regardless of opt-out settings, solely for policy enforcement purposes.
• This policy does not cover data voluntarily shared through other channels (e.g., support tickets, community forums).

7. Review

This policy shall be reviewed and updated when a new AI service is added, terms of service are materially changed, or significant changes are made to the data handling environment.

Next review date: 2026-08-27