Opt-in corrections
Correction requests are voluntary. A learner chooses what text or speech transcript to send, Aurelis runs an safety review first, and other users see only an alias, language, prompt and submitted content.
Email addresses, phone numbers, Apple identity data, subscription data and internal account identifiers are never shown in social correction surfaces.
Users can report correction requests, feedback, profiles or quests. Reported content is retained for moderation and can be hidden or reviewed.