AI Privacy & Safety
Student Safety
& Monitoring
How AI helps ensure a safe and engaging learning environment — with teachers always in control.
AI should extend teacher oversight — not replace it.
Every safeguard we build is designed to put more information in a teacher's hands, not to automate decisions about students. Safety flags, disengagement alerts, and jailbreak protections all exist to surface what a teacher needs to know — so the human in the room can make the call. AI monitors. Teachers act.
"AI monitors. Teachers act."
— The principle behind every safety feature we build
How It Works
Four layers of
student protection.
AI Content Moderation
Our AI system continuously monitors chat interactions for content that may be inappropriate, harmful, or indicative of bullying, self-harm, or other safety concerns. When such content is detected, a safety flag is applied to the specific message and the chat session — allowing teachers to quickly identify and address potentially sensitive situations. All flagged content is reviewed by the teacher and is never shared publicly.
Disengagement Detection
To support student well-being and engagement, our AI identifies potential signs of disengagement. This includes patterns such as multiple short or unthoughtful messages, a low overall message count in a session, or keywords that indicate frustration or confusion. When disengagement is detected, a flag alerts the teacher so they can proactively intervene and provide support.
Prompt Injection & Jailbreak Protection
Students sometimes attempt to manipulate an AI into behaving outside its intended scope — through roleplay framing, false authority claims, or requests to simulate "developer mode". Secondary AI has explicit, hard-coded defences against all of these techniques baked directly into every chatbot's system prompt.
Teacher Review & Oversight
Both safety and disengagement flags are designed as tools to empower teachers, not to penalize students. These flags provide valuable insights to educators, enabling them to offer timely and targeted support. Teachers have full access to review flagged content and sessions, understand the context, and make informed decisions about student well-being and academic progress.
Important Note
Flags inform. They don't penalize.
Every safety and disengagement flag is a signal to a teacher — not an automated action against a student. No flag results in any consequence for a student without a teacher reviewing it first. We designed it this way deliberately: the goal is to give teachers better visibility, not to create a surveillance system that operates without human judgment.
Learn More About Our Privacy Practices
Explore how we handle data, compliance, and responsible AI across the platform.
Privacy & Safety Overview