AI is named and explained throughout
2026-02-12
AITransparency
Buttons now say "AI" instead of "magic." Each AI feature shows which model is running (e.g. GPT Image for pictures, Haiku 4.5 for grading). If the AI changes a picture request for safety, WildReader shows what changed instead of hiding it. The parent guide and privacy policy now list all AI services used.
Builder note: Three commits in one morning: renamed magic→AI in UI, centralized model config in lib/ai-models.ts for consistent attribution, and made prompt sanitization edits visible instead of euphemistic.
Terms we used
Prompt sanitizationPractice
A safety filter that adjusts a picture request before sending it to the AI. Now you can see when it happens.
Model attributionPractice
Saying which AI model produced a result, so you know what's doing the work.