For years the “AI ethicist” was a self-attributed title, without verifiable contours. Now the picture changes: within CEN-CLC/JTC 21, EN 18274 — the European standard on competence requirements for professional AI ethicists — is on the home straight, while professional associations such as the Association of AI Ethicists work towards recognition of the profession. Who certifies, and how, those who watch over the ethics of machines?
That the ethics of AI cannot be reduced to a declaratory exercise we have long maintained: from principles for human-centred AI to their translation into verifiable requirements, the decisive step is always the one from statements to the competences of those who must implement them. It is exactly on this ridge that a notable novelty of the European standardisation site sits: EN 18274, “Competence requirements for professional AI ethicists”, developed within CEN-CLC/JTC 21 and reaching, after the public enquiry on prEN 18274:2025, the Final Draft European Standard stage (FprEN 18274:2026). The author, who takes part in the standardisation work and is a member of the Association of AI Ethicists, considers this the right place to take stock, on the basis of publicly available information alone.
What the standard provides: knowledge, skills, attitudes
According to the document’s public presentation, the standard “provides a systematized framework for the competencies of AI ethicists, categorizing them into knowledge, skills and attitudes related to the specific activities and tasks of the role, and identifies requirements and recommendations necessary for individuals to effectively perform as AI ethicists”. Two traits deserve emphasis. The first is the European values anchoring: the required competences include “a strong understanding of European values and fundamental rights” — the same anchoring that informs the AI Act. The second is the enabling, non-prescriptive nature: the document “does not impose mandatory regulations but rather provides a structured reference for developing training programs and professional certification schemes”, with the stated aim of supporting organisations “in defining roles and selecting qualified personnel”.
Why a standard about people
EN 18274 belongs to a different family from management-system or product standards: it is a standard about people. Its natural function is to ground training paths and schemes for the certification of professional competences, according to the consolidated accreditation architecture: competence requirements defined by the technical standard, bodies that assess and certify persons’ competences under ISO/IEC 17024, the accreditation body guaranteeing the reliability of the system. This is crucial for the market: when the scheme is mature, the “AI ethicist” qualification can cease to be a reputational label and become a qualification verified by an independent third party, with entry, maintenance and updating requirements. Nor should the link with binding law be neglected: in the AI Act the human dimension of governance is pervasive, from human oversight under Article 14 to AI literacy under Article 4, to the fundamental-rights impact assessment under Article 27 of the Regulation — all functions presupposing structured ethical-legal competences.
The associative front: the Association of AI Ethicists
Alongside technical standardisation runs the associative movement. The Association of AI Ethicists (AAIE), based in France, declares its mission thus: “the Association of AI Ethicists exists to promote and protect the professional interests of AI ethicists and other professionals working in the field of ethical AI”. Significant, for our purposes, is the commitment on qualification: the association declares it develops and promotes “a robust certification programme for professional AI Ethicists that provides individuals with a recognized standard of expertise”, also “becoming a direct liaison with standardisation bodies as ISO/IEC and CEN-CENELEC”. The convergence between the associative and normative fronts is the clearest signal of the maturation under way.
Critical issues
Three knots remain open. First, the risk of a returning “ethics washing”: the availability of a certified qualification must not turn the ethicist into a stamp of moral conformity affixed downstream of decisions already taken; the organisational placement and independence of the function count as much as the competences. Second, the relationship with regulated professions: the AI ethicist’s activities intersect legal assessments that, in national systems, belong to chartered professions, and the boundary will have to be carefully overseen. Finally, effectiveness: a competence standard lives on serious certification schemes, competent bodies and aware market demand.
Conclusions
The parabola is recognisable: as already for auditors and privacy professionals, so for AI ethics the credibility of the function passes through the verifiability of competences. EN 18274, once published, will offer Europe the common vocabulary; it will fall to associations, organisations and certification bodies to turn it into practice. In the light of the above, one wonders whether the market will welcome this professionalisation without hollowing it out, giving rise to serious qualification schemes, and not to a mere issuing of certificates, so as to offer organisations what the AI era most needs: competent, independent and responsible people safeguarding human dignity in the decisions of machines.