Universities bring AI onto campus as EU rules tighten oversight
Universities are moving from improvised classroom rules on generative AI to institution-wide contracts, policies and assessment redesign. Somdeep Sen’s 11 June opinion argues that cash-strapped campuses risk treating students as an AI labour pipeline; the wider evidence shows a more mixed but urgent governance problem. OpenAI says ChatGPT Edu is built for universities, with privacy controls and higher usage limits. California State University announced a systemwide rollout in 2025, while faculty objections at the University of Colorado led the system to delay student access until autumn 2026. For Belgium, the question is not whether students use AI but how KU Leuven, UCLouvain, ULB, VUB, UGent and university colleges govern it under EU rules. The European Commission says AI used for exam scoring or access to education can be high-risk, while AI literacy duties already apply across the EU.
This matters first to students, parents, lecturers and university administrators in Belgium. Belgian campuses already face the same classroom questions: when AI help is allowed, how assessment remains fair, and whether paid institutional tools widen or narrow access. The European Commission says AI literacy duties already apply, and the AI Act treats some education uses as high-risk. Belgian universities and university colleges therefore need policies that protect learning, privacy and equal treatment rather than simply buying campuswide tools.
Somdeep Sen (academic at the University of Pretoria in South Africa) wrote the lead opinion. Cisco (US networking and technology company) has promoted AI-enabled efficiency in higher education. OpenAI (US AI company behind ChatGPT and ChatGPT Edu) sells university-focused AI access. California State University (23-campus US public university system) became a prominent test case for campuswide AI deployment. Glendale Community College (Arizona college) drew attention after an AI name-reading system failed at graduation. University of Cambridge (UK research university) led recent work on AI and essay assessment. Deborah Talmi (Cambridge psychologist) is cited in that assessment debate. University of Minnesota, Dartmouth College and Syracuse University (US higher-education institutions) are examples of universities entering AI partnerships. The European Commission (EU executive in Brussels) oversees implementation of the AI Act through the European AI Office.
Background
The current debate follows the November 2022 release of ChatGPT, when universities first responded with bans, plagiarism warnings and improvised classroom rules. UNESCO’s 2023 guidance called for privacy protection, human-centred use and coherent policy frameworks in education. The European Commission’s 2022 educator guidelines had already framed AI literacy as a teaching and governance issue. The AI Act entered into force on 1 August 2024, with AI literacy obligations applying from 2 February 2025 and high-risk education rules now scheduled by the Commission for 2 December 2027 after simplification talks.
The wider picture
AI in universities sits inside a broader competition over who controls AI infrastructure, talent pipelines and standards. US firms supply many tools, while the EU is trying to combine adoption with rights-based regulation. Universities are strategically valuable because they train future workers, produce research and legitimise technologies that companies want embedded across public life.
Why now
The 11 June 2026 lead revives the issue as university AI contracts and faculty pushback become more visible. It also lands after AI literacy duties began applying in the EU and as the Commission prepares the next stages of AI Act implementation for education-related systems.
What to watch
Watch whether Belgian universities publish clearer AI assessment rules before the 2026-27 academic year, whether community education authorities issue common guidance, and whether EU implementation tools clarify what counts as high-risk AI in exams, admissions and student evaluation.
Impact
Regional — The EU sets the common regulatory floor through the AI Act, including literacy duties and future obligations for high-risk education systems. Belgium’s higher-education choices sit mainly with the Flemish Community and French Community, with Brussels campuses often operating in both language systems. That means KU Leuven, UGent, VUB, ULB, UCLouvain and university colleges may face the same EU compliance framework but different community-level guidance, procurement routes and assessment cultures. Federal Belgium is less central except for data protection and labour-market spillovers.
Opposing perspectives
- University technology adopters
OpenAI says ChatGPT Edu gives universities privacy controls, administrative settings and campuswide access that individual consumer accounts do not provide. The strongest pro-adoption case is that students are already using AI, so universities should provide safer, more equal access while teaching responsible use.
- Academic governance and faculty voices
Faculty objections at the University of Colorado centred on classroom disruption, privacy, bias, hallucinations, environmental footprint and the risk that contracts are signed before academic governance catches up. Their strongest argument is not a blanket anti-AI stance but that assessment rules and teaching autonomy should precede procurement.
- Human-centred education advocates
UNESCO’s 2023 guidance says generative AI policy should protect privacy, validate tools and preserve a human-centred vision of education. That frame treats AI as potentially useful but rejects a model where efficiency, automation and labour-market alignment become the university’s organising purpose.
- Critical university scholars
Somdeep Sen argues that universities under financial pressure risk accepting a corporate framing in which students become a supply chain for AI-ready labour. The strongest version of this view is that universities should teach AI literacy while defending mentorship, judgement and critical thought as public goods.
Sources & evidence
- View sourceAl Jazeera - The university must not become a supply chain for AIPrimary· aljazeera.com· 11 June 2026Retrieved 11 June 2026· 34 days ago· Dated
- View sourceEuropean Commission - AI Act· digital-strategy.ec.europa.euRetrieved 11 June 2026
- View sourceEuropean Commission - Ethical guidelines on AI and data in teaching and learning· education.ec.europa.eu· 25 October 2022Retrieved 11 June 2026· 1359 days ago· Dated
- View sourceUNESCO - Guidance for generative AI in education and research, Fengchun Miao and Wayne Holmes, 2023· unesco.org· 7 September 2023Retrieved 11 June 2026· 1042 days ago· Dated
