This board operationalises a lot of existing and forthcoming standards, frameworks, policy and guidance. It provides concrete indications of how broader responsible and ethical AI use could look in practice. It isn't, however, a certification or a legal judgement on a tool.
Read the source: EU AI Act, official text on EUR-Lex. Links open in a new tab, so you can check any quote on this page for yourself.
“AI systems intended to be used to evaluate learning outcomes, including when those outcomes are used to steer the learning process” Annex III (opens the source document in a new tab)
Tools that do these things are the board's core subject. The more of a pupil's learning a tool steers, the more checks apply to it.
“remain aware of the possible tendency of automatically relying or over-relying on the output… able to intervene in the operation of the high-risk AI system” Article 14(4) (opens the source document in a new tab)
Article 14(4) lists what oversight must include: understanding the system, watching for over-reliance, interpreting its output, deciding not to use it, and intervening or stopping it. Each of these is a separate check on this board.
“the right to obtain from the deployer clear and meaningful explanations of the role of the AI system in the decision-making procedure” Article 86(1) (opens the source document in a new tab)
The board scores whether an explanation is given at the time of the decision, and whether it can be challenged. The second part goes further than the Act.
“examination in view of possible biases that are likely to affect the health and safety of persons, have a negative impact on fundamental rights” Article 10(2)(f) (opens the source document in a new tab)
The board's fairness area is this examination, done on the tool's public record: recurring checks, broken down by pupil group, shared with schools and families.
“prohibits AI systems to infer emotions of a natural person in the areas of workplace and education institutions” Article 5(1)(f), in force now (opens the source document in a new tab)
The board records any profiling or emotion claims found in a tool's public documentation.
Bans and AI-disclosure duties bind now; the full high-risk duties for education tools arrive December 2027. EU AI Act (opens the source document in a new tab)
Vendors have until December 2027 to meet the high-risk duties. The board shows today which of them already do.
Schools that use high-risk AI are deployers under the Act, with duties to use systems properly and assign oversight to people with the competence and authority to exercise it. Article 26 (opens the source document in a new tab)
The board gives schools the evidence those duties ask for, without each school repeating the work.