Draft for attorney review. This page is not final legal advice. The owner will have a qualified attorney review it before public launch. Last updated 2026-10-19.
Where AI is used
- Aura Score: computer-vision models on our own servers measure facial landmarks, geometry, presentation, and photo quality, and combine them into a range with a confidence level.
- Roasts and explanations: Anthropic's Claude Sonnet writes them from your scan's numbers. It never sees your photo.
- Deep Aura Report: Claude Opus writes a longer breakdown from the same numbers.
- Moderation: Claude Haiku screens posts, comments, bios, and messages for safety.
Limits you should know about
- Scores are estimates with real uncertainty, which is why we show a range. Lighting, angle, filters, and expression change them.
- Models can reflect bias in the data they learned from. We review flagged results for bias and adjust the scoring, but we can't promise every result is fair.
- Roasts are written to be "savage but safe": never about race, ethnicity, skin condition, disability, weight, age, gender identity, religion, or health. Report any that cross the line.
- AI moderation makes mistakes in both directions; held content goes to a human.
What AI doesn't decide
AI doesn't decide bans, refunds, or anything about your account beyond holding content for review. Aura Points come from completing regimens, never from your score.