LEGAL
AI Usage Policy
How Lunxa AI works, which providers are involved, its limitations and your responsibilities.
- Version:
- 0.9-draft
- Last updated:
- November 7, 2026
1. Purpose
Lunxa AI helps you understand information, draft content, summarise documents, plan and organise work, and find knowledge inside your workspace.
2. Providers
Requests are processed by the AI provider(s) configured by Lunxa — currently Anthropic and/or OpenAI. Your request and the minimum relevant context are sent to generate the response. [Provider data-retention and training terms to be confirmed against active contracts.]
3. Your responsibilities
Use AI features lawfully and in line with our Acceptable Use Policy. Don't input data you aren't permitted to process. You are responsible for how you use AI output.
4. Accuracy
AI output may be inaccurate, incomplete, biased or out of date. It is labelled as AI-generated and is not verified human work. Check important information against its sources.
5. Human oversight
AI can propose actions (such as creating tasks). Consequential actions require your explicit confirmation, and applied AI actions can be undone.
6. Data handling
Prompts, responses and feedback are stored in your workspace so you can revisit conversations; you can delete conversations at any time. Usage metadata (tokens, cost, latency) is kept for billing, limits and capacity planning. Lunxa staff don't see your prompts by default.
7. Permission boundaries
Lunxa AI uses your permissions. It cannot access other workspaces, documents you can't open, or documents with AI access switched off. Revoked access takes effect immediately.
8. Prohibited uses
Generating unlawful, deceptive or harmful content; attempting to extract other users' data; circumventing limits; using AI output to make decisions with legal or similarly significant effects on people without human review.
9. Automated action safeguards
AI never silently deletes data or sends external communications. Automations run only what their owner configured, with run history and limits.
10. Training and retention
Lunxa does not train its own models on your content. Provider practices are described in section 2.
11. Feedback
Use the thumbs up/down controls on any response, or contact support, to report problems with AI output.
12. Changes
We'll update this policy as AI functionality changes and notify you of material changes.