Security, privacy and voice consent

Your content

• Uploaded PDFs, videos, slides and question banks are stored for your account only and used solely to compile and teach your lessons.
• Compiled teaching plans are derived works of your content and belong to you. You can preview every board before a learner sees it and recompile or remove any of it.
• We do not use customer content to train or fine-tune models, ours or anyone else’s.
• On account closure, content and compiled plans are deleted on request.

Teacher voice and likeness

This is the part US customers rightly ask about first, so the rules are strict and simple:

1. Consent is explicit and recorded. A voice is cloned only from a sample recorded or uploaded by the teacher or account holder themselves, after they accept a consent statement in the product. We keep the consent record with the voice.
2. We do not clone third parties. No public figures, no colleagues from found audio, no “can you make it sound like…”. Avatars are built only from a photo the person provides of themselves.
3. Private to your account. Nobody else on the platform can select your teacher’s voice or avatar.
4. Disable or delete at any time. Disabling in Tutor Mode settings switches lessons to a stock voice immediately. Deletion removes the voice model and the sample.
5. Stock voices need no consent. If you would rather not clone anyone, the stock voice set (including American English) is included at no charge.

Student and learner data

• We store what the product needs to teach and report: position in the course, answers to checks, mastery per concept, session transcripts, notes, and minutes used.
• Transcripts and reports are visible only to admins of your account. They are exportable (CSV) and deletable on request.
• We do not sell or share learner data with advertisers or data brokers.
• If your learners are minors, you are the party with the relationship to parents and school; tell us on the call and we will document the data flows for your records.

What the AI can and cannot do

• The whiteboard is drawn from a compiled, reviewable plan using a fixed set of board operations. The live model never authors raw page content.
• Narration is compiled ahead of time; the live model adapts pace and answers doubts against the material on the board.
• The live model is used for grading answers, giving hints, answering doubts and adapting to the learner. It does not browse the internet or pull from other customers’ content.
• Every session is logged, so an admin can read exactly what was said to a learner.

Infrastructure and access

• Data is encrypted in transit. Production access is limited to the engineers who operate the service.
• Language and speech models are provided by third-party AI providers under contract; content is sent to them only to perform the requested task and not retained by us for any other purpose. The current provider list and hosting region are available on request.
• We do not currently hold SOC 2 or ISO 27001. We will say so on this page until we do.

Your rights and how to exercise them

• Export your data: from the admin, or ask us and we send a CSV bundle.
• Delete a voice, an avatar, a learner’s records, or the whole account: email [email protected] from an admin address; we confirm in writing when it is done.
• Report a security concern: same address, marked “security”. We answer within one business day.

This page describes the product as of September 2026 and is updated when anything changes. The Vacademy privacy policy is the legal document behind it.