AI Transparency
Working draft — pending review by qualified Indian legal counsel. It is not yet a binding policy, and highlighted [PLACEHOLDERS] will be filled before launch.
Where AI may help, what it never decides, and how to ask for human review.
AI features are currently switched off on AstroTrust. Every verification, complaint, moderation and appeal decision is made by a person.
If AI assistance is enabled after review, it will only help staff — for example by suggesting a complaint category, spotting duplicate text or drafting a neutral summary. Every AI output is marked as a suggestion and needs human acceptance. AI can make mistakes.
AI will never
- Declare anyone fraudulent, dishonest, unsafe, guilty or unfit.
- Approve or reject identity or credentials, decide whether a complaint is true, publish allegations or suspend an account.
- Judge whether a prediction is accurate.
- Infer religion, caste, health or other sensitive traits.
- Receive identity documents, complaint evidence or reporter details without documented necessity, contractual safeguards and legal review.
Your rights
You can ask for human review of anything AI assisted with: [SUPPORT EMAIL].