Generative artificial intelligence (GenAI) is rapidly becoming embedded in students’ study practices, challenging universities to balance innovation with academic integrity, equity, and valid assessment. For the CDIO implementation community, the key question is which institutional mechanisms help programmes and teachers integrate GenAI while protecting intended learning outcomes and assessment credibility. We analyze GenAI rules and guidelines at six Nordic CDIO institutions using Diffusion of Innovations Theory (compatibility, trialability, observability). Key findings: (1) baseline principles converge on student/teacher accountability, growing expectations for disclosure, and GDPR-driven steering toward sanctioned tools (e.g., Microsoft Copilot); (2) trialability is operationalized through implementable “rules-in-use” (traffic-light models, course templates, declarations) that translate policy into course practice and support CDIO Standards 9 and 11; and (3) observability varies with organizational context: multi-faculty universities depend on recurring cross-faculty partners (central pedagogical units, AI labs/policy labs, working groups) to diffuse and update guidance, while more centralized or single-faculty settings emphasize streamlined recommendations and teacher-led, course-specific implementation. Implementation implications for CDIO: (i) adopt shared course-level decision artefacts to enable consistent yet discipline-sensitive implementation; (ii) resource faculty support for assessment redesign (Standard 9); and (iii) build feedback loops for continuous improvement and assessment transparency (Standard 11).
