Safety
A space for students, with schools involved from the start. This is our planned approach; the student platform and its safety systems are still in development.
Designed around schools
Only invited schools
We vet every partner school. No stranger can register.
School sign-in
Students sign in with the Google account their school gives them. The school already knows who its students are, so we don't ask for ID or scan faces.
No adults besides moderators
Only students and moderators approved by their school can use the platform.
Minimal account details
The planned account details are a school email (kept hidden), a username, an optional profile photo, school name, country, grade and age group. Each student's school sets their age group, so we never ask for a date of birth. The platform will also need to process posts (including photos, which can show faces), comments, private messages, projects and moderation records. Data handling and retention details are still being finalised before students join.
Usernames only
Students appear as a username with their school and country, including in the public showcase. No full names.
Private spaces, not spying
Students can message each other and create private spaces, and talk normally there. They can send pictures and stickers there, but never links or videos, and chats are just for students in the same group: Prodigies (10–12), Rangers (13–15) or Visionaries (16+). Everyone mixes freely in clubs and projects. No adult reads those conversations by default. An AI checks the words regularly and tells the school's moderator if a conversation breaks the rules. The AI can't see pictures, so every picture has a Report button.
Photos
Students can share photos in their posts and projects, and yes, faces are fine. Every photo goes through the same automatic check, and your school's mod decides if one crosses a line for your school. No videos.
Community norms
Rules are written as what keeps this a good place for everyone, not as a list of bans.
Moderation
AI checks every post, comment and private conversation at the end of each day, and flags anything really serious straight away. It reads private conversations in full, so it can spot a problem that builds up over time. It looks at meaning and context, not just a list of banned words.
Swear words are blurred automatically in the feed, comments and private chats, and students can't reveal them. A moderator sees the real words only when opening a tier 3 or 4 case.
The AI reads words, not pictures. When a student reports a picture in a private chat, the moderators at the schools of the students involved can open that one picture, and nothing else from the chat. In tier 3 and 4 cases they can open the pictures in the conversation too.
Each partner school names a moderator. They look after their own school's students, and help look after up to five clubs they choose. Moderators never browse conversations. For smaller problems they get the AI's report, not the messages. For serious ones they're told right away and can open the messages involved.
Mods have a few simple tools: pause someone's chats for a while, set quiet hours for one student or the whole school, or lock an account until they've had a chat.
Routine moderation is initially estimated at 15 to 30 minutes a week; incidents and project approval may take longer.
Tier 1: minor
Normal kid behaviour, sarcasm, borderline jokes.
Nothing interrupts the student. The AI adds it to a weekly summary for the school's moderator, who usually takes no action.
Tier 2: clear but moderate
A clear break of the community norms.
After the end-of-day check, the moderator gets an alert with the AI's report and the student's history, but not the private messages. They decide what happens.
Tier 3: repeated or more serious
The same problem again, or something worse.
The moderator is notified right away and can open the messages involved. The consequence is stronger, such as a longer restriction or a talk with the student.
Tier 4: serious
Real threats, sexual content, signs of self-harm or grooming.
The post or message disappears and the account locks straight away, without waiting for anyone. The mod is told at the same moment. If it turns out to be a false alarm, they can bring it all back.
Consequences fit the problem, and only limit what was misused. A moderate mistake doesn't remove a student from their clubs or projects. Only tier 4 affects the whole account.