By design, Brian aligns with the EU AI Act: it is a GDPR/DSG-compliant AI tutor hosted on European infrastructure with strict data minimization and 'no training on customer data' principles. It also has an Acceptable Use Policy that enforces privacy by design and the safe and lawful use of AI.
In practice, Brian functions as a support tool for teaching and learning. It helps to generate and adapt exercises, provide formative feedback and surface insights for teachers. Teachers remain in full control of curricula, grading, promotion and access decisions.
As Brian's outputs are not used autonomously to make decisions such as admission, grading, certification or other high-stakes educational choices, it is classified as a limited risk AI system under the EU AI Act's education category.
Brian's architecture and pedagogy (transparent AI use, human-in-the-loop design and robust data protection safeguards) already implement the core principles of the AI Act.