Governing Artificial Intelligence in Conformity Assessment: The ISO/CASCO Perspective
Governing Artificial Intelligence in Conformity Assessment: The ISO/CASCO Perspective
Publish Date: 2026-02-22 00:00:00
Source Domain: www.qualitymag.com
Artificial intelligence is often portrayed as a disruptive force that standards and accreditation systems must urgently “catch up with.” In reality, within the ISO/CASCO(1) ecosystem, AI is neither uncharted nor ungoverned. It is already present, already anticipated, and already subject to well-established conformity assessment principles.
The ISO/CASCO framework has always been deliberately technology neutral. It regulates outcomes, responsibilities, competence, impartiality, and trust. This design choice has proven prescient. As AI-enabled tools enter certification, inspection, and scheme management, CASCO standards already provide a robust governance structure without needing to chase technological trends.
Recent revisions and drafts within the ISO/CASCO ISO/IEC 17000 series increasingly acknowledge the use of digital and automated tools, including artificial intelligence (AI)(2)(3), within conformity assessment and accreditation activities. AI is considered critical whenever it affects any part of the selection, determination, review, decision, attestation, surveillance, or acceptance of results.
Standards such as ISO/IEC 17024 (personnel certification), ISO/IEC 17067 (certification schemes), and ISO/IEC 17020 (inspection bodies) now explicitly or implicitly address algorithm-supported processes, remote and automated evaluations, data-driven decision support, and technology-enabled conformity assessment functions.
Explicit AI governance in CASCO standards
The most direct and mature treatment of AI appears in ISO/IEC FDIS 17024:2025 for certification of persons. For the first time, CASCO explicitly defines artificial intelligence and permits its use (for example, in examination invigilation), while simultaneously imposing strict conditions. Where AI is used in any certification activity, the certification body must demonstrate control of impartiality risks (including AI related bias), ensure human oversight, validate AI-supported outcomes, demonstrate…
