Keynote at the T4E Transformational Leadership Programme, held 13-14 May 2026 in the University of Alicante, Spain. This keynote argues that universities must move beyond viewing artificial intelligence as a mere technological trend and recognise it as a core challenge for institutional governance, digital transformation, and academic responsibility. The presentation begins by framing digital transformation and AI as key terms in higher education government, but immediately questions a technology-centred view. Digital transformation is presented not only as the optimisation of processes through technology, but as a change in mindset, operating models, and institutional culture. Its central element is people, not tools. The talk then defines the real challenge for universities: rethinking digital transformation from digitising processes to governing AI-enabled sociotechnical ecosystems with meaningful human oversight. AI is shown as affecting the three main university functions: teaching, administration, and research. In teaching, it enables personalised learning and engagement; in administration, automation and efficiency; and in research, data analysis and discovery acceleration. However, the presentation stresses that AI also creates risks: bias, opacity, legal non-compliance, privacy breaches, academic integrity concerns, dependence on third-party providers, and uneven access. A major section focuses on responsible AI adoption through the Safe AI in Education Manifesto, whose principles include human oversight, confidentiality, alignment with educational strategies and didactic practices, accuracy, explainability, comprehensible interfaces, ethical model training, and transparency. These principles map to university governance strategies: human-oriented, infrastructure-oriented, and a governance/assurance layer. The presentation also highlights the need for ethical AI policies, critical AI literacy, communities of practice, and shared good practices. The keynote further explores the strategic dilemma between relying on third-party proprietary tools and developing one’s own infrastructure based on open LLMs. It argues that there is no single best option: universities must evaluate privacy, cost, internal capacity, transparency, auditability, deployment speed, and strategic autonomy. In-house intelligent applications, such as learning assistants, are presented as examples of governed institutional services. The closing message is that AI governance must be strategic, participatory, and ethical. Universities should not merely adopt AI systems but build an AI-augmented academic culture grounded in values, critical engagement, institutional responsibility, and human-centred innovation.