Responsible AI Principles

We optimise for understanding, curiosity and capability — never for time-on-screen, streaks or compulsive engagement. If a feature makes a child more dependent on a screen without making them more capable, it does not ship.

AI in UPSTYE products is designed to prompt thinking, scaffold problem-solving and adapt pace — not to replace the child’s own reasoning, a teacher’s judgement, or a parent’s role. Answers are earned, not dispensed.

AI features operate within clearly defined educational boundaries appropriate to the product and age group. We constrain what the AI can discuss, generate and access. Open-ended, unsupervised AI conversation is not an appropriate default for children, and it is not ours.

We will always state clearly, in plain language:

• Where AI is used in a product and what it does.
• What data the AI feature uses and where it is processed.
• What the AI cannot do, and its known limitations.
• How parents can view, limit or disable AI features.

AI features are designed to work with the least personal data possible. We do not build advertising profiles of children, we do not sell learner data, and we favour on-device processing wherever technically feasible.

Educational content generated or adapted by AI is subject to human review processes. Feedback and error-reporting channels are provided, and reported issues are investigated by people, not just models.

AI behaviour, vocabulary, content boundaries and safety filters are tuned to the stated age range of each product — and default to the most protective setting.

We will not market AI capabilities we have not built, outcomes we cannot substantiate, or “personalisation” that does not exist. Development-stage features are labelled as such.

These principles apply to every UPSTYE product that uses AI, and they are reviewed as our products and the regulatory landscape evolve. If you believe a product falls short of them, write to info@upstyeeducation.com — it will be read.