AI Classroom Privacy Policy

AI Classroom is built and operated by Chinland InfoTech, which is responsible for this policy. Where a school or programme runs AI Classroom for its own learners, that organisation decides who takes part and what is recorded, and we run the platform on its instructions — in that case its own notice governs those decisions, and it is the right first contact about a learner's records.

• Your voice, during the lesson. What you say is carried to the service so it can be recognised and answered. It is turned into text as it arrives, and unless recording has been consented to, the audio is not written anywhere.
• Lesson progress. Which lesson you are in, how far through you are, what you answered and whether you completed it — so the lesson can adapt and so your organisation can see progress.
• A session. A learner session identifier tied to your organisation, which expires on a rolling thirty days.
• Access and security records. Sign-in attempts and rate-limit counters, kept to protect the service from abuse.

Keeping a learner's voice is the exception and is built that way rather than promised that way.

• A consent token is required. Every retained clip records the consent it was kept under. Without one, no clip is written.
• It is stored apart from the lesson service. Retained audio and its transcript live in a separate area of the database that the application running your lesson has no permission to read. Reaching it takes a different, restricted role.
• Every clip has an expiry. A retention date is set when the clip is stored, and a sweep deletes both the audio file and its database row once that date passes. This is enforced by the system, not by someone remembering.
• What it is for. Retained clips exist to check and improve recognition — particularly for Burmese and Falam Chin, where usable training material barely exists. They are not sold, not shared, and not used for advertising.

Where several organisations use AI Classroom, each one's records are isolated at the database level rather than by application code being careful — a query made for one organisation cannot return another's rows. Lessons, learners, sessions and branding all belong to a single organisation.

Live audio and video for a lesson travel over a real-time transport service. Speech recognition and the models that teach the lesson run on infrastructure we operate. We use cloud hosting, database and object storage providers to run the platform, each acting on our behalf.

We do not sell anything a learner says, and we do not share it with advertisers or data brokers. We disclose information only where the law requires it, or where it is necessary to protect the security of the service.

AI Classroom is intended to be used by learners, including children, through a school or programme rather than by a child signing themselves up. Where a child takes part, the organisation is responsible for obtaining consent from a parent or guardian — including consent to keep audio, if it chooses to enable that — and for deciding whether keeping recordings is appropriate at all for its learners.

If you believe a child's recording is held and should not be, contact the organisation running the lesson, or us, and we will act on it.

Taking part in a lesson requires that your speech reaches the service; there is no version of a voice-led lesson that avoids it. Keeping a recording afterwards is separate, and is refusable without losing the lesson.

To ask what is held, to have a recording deleted before its expiry, or to raise a privacy concern, contact your school or programme, or email admin@chinlandinfotech.com. Your rights may vary by location, and we may need to verify a request before acting on it.

Traffic travels over encrypted connections. Access to retained audio is restricted to a database role separate from the one the lesson service uses, so a fault in the lesson application does not expose recordings. Sessions expire on their own, and repeated failed sign-ins are rate-limited.

AI Classroom is still in development, and the live voice stack is not yet finished. This page describes how the system is built to behave; it will be revised as the product reaches learners, and the effective date above updated.