Recording consent

What this is about

A child can answer by speaking. The microphone records their voice for a short time. The recording is turned into words. The words are used to give feedback.

Voice is optional. A child can always type instead.

Who agrees

A parent or guardian agrees for a child under 15. They choose it when they create the child's account. If they do not tick it, the child types.

What happens to the recording

• The browser sends the audio to Microsoft Azure Speech in Australia, through our server.
• Azure turns it into words. Microsoft says real-time speech-to-text does not store the audio.
• We use the recording only to turn it into words.
• We delete our copy within 24 hours by default. A parent can choose a longer time in settings.
• The words are kept as part of the child's answer. They go to Anthropic in the United States, with names hidden, so the AI can write feedback.

Some first tries are private

On tasks that offer one, a child's first try is private, even if they speak it. The audio still goes to be turned into words, so the child can check them. Then the audio is deleted. The words are removed when the lesson ends, or 24 hours after it was left open. Nobody else sees them.

Please know

• An adult sits nearby during practice.
• The words can be wrong. A child can fix them.
• A voice can say things about a person. Please remind your child not to say private details like a last name, address or school.
• We try to hide names and addresses, but this can miss some.

Changing your mind

A parent can turn voice off at any time. The child then types instead. A parent can also withdraw consent for the child, or delete data. In the app: Settings, then the child. Or write to the contact address in the app.

Turning voice off stops new recordings. The child can keep practising by typing. Withdrawing consent for the child stops new lessons until a parent agrees again.