Interview With a University President: How We Govern AI in a Spanish Public University

Spanish public universities are deploying AI systems at an accelerating pace—from automated grading assistants to predictive analytics for student dropout—yet many lack formal governance frameworks. In an exclusive interview, the president of a mid-sized Spanish public university shares how her institution built an AI ethics committee, navigated faculty resistance, and drafted enforceable policies for generative AI in classrooms. The approach offers a blueprint for institutions grappling with similar challenges across Europe and Latin America.

This conversation matters because governance gaps create legal and ethical risks: discriminatory algorithmic decisions, data breaches, and academic integrity crises. As the European AI Act enters force in 2026, universities must move from pilot projects to accountable, transparent systems.

Context: The Governance Vacuum in Spanish Higher Education

According to a 2025 survey by the Conference of Spanish University Rectors (CRUE), 68% of Spanish public universities use at least one AI system in academic or administrative processes, yet only 22% have published governance policies. The gap reflects broader inertia: Spain’s higher education sector operates under 1980s-era organic laws, predating the internet. Regional autonomy fragments oversight, and budget constraints delay digital transformation investments.

Meanwhile, the European AI Act classifies certain university applications—automated admissions decisions, exam proctoring software with biometric recognition—as «high-risk» systems requiring conformity assessments, transparency documentation, and human oversight. Universities face penalties of up to 6% of annual revenue for non-compliance starting in mid-2026. The regulatory deadline forced many institutions to improvise governance structures in months rather than years.

Our interviewee, whom we’ll call Dr. Elena Ruiz (a pseudonym to protect institutional confidentiality during ongoing policy negotiations), presides over a university with 18,000 students across sciences, humanities, and engineering. She spoke with StudyVerso in March 2026, two weeks after her institution’s AI governance framework received final approval from the university senate.

Building the Ethics Committee: Composition and Power

Dr. Ruiz’s first decision was structural: the AI ethics committee would hold veto power over departmental deployments, not merely advisory status. «We saw too many cases where a well-meaning professor would pilot a grading algorithm without checking for bias or data protection compliance,» she explains. «Advisory committees get ignored. We needed enforcement teeth.»

«The committee has the authority to halt any AI project—including those funded externally—if it identifies unmitigated risks to students’ rights or academic integrity.»

The committee comprises seven members: two faculty (one STEM, one humanities), one PhD student representative, the data protection officer, an external legal expert specializing in algorithmic accountability, an external AI auditor, and a rotating seat for student union nominees. Terms last two years with staggered renewals to preserve institutional memory.

Critically, the committee operates with a published rulebook. Any department planning to use AI must submit a mandatory impact assessment at least 60 days before deployment. The assessment checklist includes data sources, algorithmic logic (if proprietary, a third-party audit suffices), accuracy metrics, bias testing results, and a human oversight protocol. The committee reviews within 30 days and can request modifications, impose monitoring conditions, or reject outright. Decisions and anonymized assessments are published online, creating a searchable precedent library.

This transparency proved contentious. Some faculty argued that publishing impact assessments would expose competitive research advantages. Dr. Ruiz countered that public accountability outweighed those concerns, especially when students’ educational outcomes were at stake. The senate ultimately sided with transparency, though proprietary commercial details can be redacted.

Real Challenges: Faculty Resistance and Budget Constraints

Implementing governance revealed deeper institutional tensions. A notable flashpoint emerged when the engineering faculty piloted an automated plagiarism detection system using large language models to flag suspicious assignment patterns. The system generated a 40% false-positive rate during testing—tagging original work as AI-generated because students used formal academic language taught in writing workshops.

The ethics committee blocked the deployment and required the faculty to manually review all flagged cases with students present. Faculty leaders protested that the manual process was unsustainable with 600 students per semester. Dr. Ruiz’s response was unequivocal: «If you can’t resource human oversight, you can’t deploy the system. Efficiency gains cannot override due process.»

The standoff lasted three months. The engineering faculty eventually secured funding for two additional teaching assistants dedicated to review panels, and the plagiarism tool was redeployed with stricter thresholds and mandatory human arbitration. The case became a governance precedent: labor costs of oversight must be budgeted upfront, not deferred.

Budget constraints proved equally thorny. External AI audits cost €8,000–€15,000 per assessment. The university’s annual digital transformation budget is €200,000, covering infrastructure, cybersecurity, and training. Dr. Ruiz negotiated a consortium agreement with four neighboring universities to share auditor contracts and split costs, reducing per-institution expenses by 60%. The model has since attracted interest from universities in Valencia and Andalusia.

Student Rights: Opt-Out Mechanisms and Data Minimization

Student unions raised concerns about predictive analytics systems tracking attendance, library usage, and course performance to flag at-risk students for early intervention. While the university framed this as a retention tool, union representatives argued it created a surveillance infrastructure with potential for misuse—by future administrations or through data breaches.

The ethics committee mandated an opt-out mechanism. Students can exclude themselves from predictive models without penalty, though they forfeit proactive outreach (academic advisors won’t receive automated alerts about their engagement patterns). Approximately 18% of undergraduates opted out in the first semester, a figure Dr. Ruiz considers healthy. «If everyone opted out, the system would be unworkable. If no one did, we’d have failed to communicate the stakes,» she notes.

The committee also enforced data minimization principles inspired by GDPR. An initial proposal to feed predictive models with students’ social media activity, scraped via university Wi-Fi metadata, was rejected outright. Only institutional data—attendance records, grades, library checkouts—can be used, and students receive annual reports showing exactly what data points were processed.

Implications for the Broader Higher Education Sector

Dr. Ruiz’s governance model illustrates a viable path for resource-constrained public universities, but it demands political will and cross-institutional collaboration. Three lessons emerge from the case:

First, veto authority matters. Advisory ethics boards without enforcement power become performative, rubber-stamping deployments to avoid conflict. Dr. Ruiz’s committee wields real leverage because the university charter explicitly grants it decision rights, approved by the senate as a bylaw amendment.

Second, transparency and precedent reduce friction over time. The published impact assessment library now guides faculty through pre-submission compliance, cutting committee review times by half. Departments self-correct obvious issues before formal submission, learning from prior cases.

Third, cost-sharing arrangements make rigorous oversight feasible. Auditor consortia, shared legal counsel, and open-source compliance checklists democratize governance capabilities beyond elite, well-funded institutions. Dr. Ruiz is working with CRUE to standardize impact assessment templates across Spanish universities, potentially creating a national governance commons.

Challenges remain. The university has yet to tackle faculty use of generative AI in research—ChatGPT for literature reviews, image generators for grant proposals—which falls outside current committee scope. Dr. Ruiz acknowledges this is «the next frontier,» requiring consultation with research ethics boards and alignment with European research integrity guidelines.

Comparisons to other jurisdictions reveal divergent approaches. UK universities lean heavily on Quality Assurance Agency (QAA) guidance, which emphasizes principles over prescriptive rules. Spanish universities experimenting with AI tutoring systems face similar governance questions about algorithmic pedagogy and learning analytics. In the United States, decentralized governance leaves each institution to navigate AI risks independently, creating wide variance in student protections.

The European AI Act’s enforcement phase will test whether Dr. Ruiz’s model scales. If compliance audits reveal that formalized governance correlates with fewer violations and better student outcomes, other universities may adopt similar structures. If the administrative burden proves unsustainable, institutions might retreat to minimal compliance postures, risking both legal penalties and ethical harms.

What Comes Next: Generative AI in Classrooms and Research Integrity

The governance framework Dr. Ruiz built focuses primarily on administrative and analytics systems—admissions algorithms, dropout prediction, plagiarism detection—but generative AI introduces new complexities. Faculty are already using large language models to draft syllabi, generate exam questions, and provide students with AI tutoring supplements. These applications don’t fit neatly into «high-risk» regulatory categories, yet they reshape pedagogy and assessment in profound ways.

The ethics committee recently opened a consultation process on classroom AI, surveying faculty and students about current usage patterns. Preliminary results show that 34% of faculty have experimented with ChatGPT or Claude for course design, while 52% of students report using generative AI for assignments (despite ambiguous policies on permissibility). Dr. Ruiz plans to publish university-wide guidelines by September 2026, before the academic year begins.

Key questions under debate include: Should students be required to disclose AI use in assignments? How should faculty adjust learning outcomes if AI makes certain skills obsolete? What constitutes academic misconduct when AI is a legitimate study aid in professional contexts? Degrees with strong tech industry connections face particular pressure to integrate AI literacy into curricula, blurring lines between tool use and shortcut.

Dr. Ruiz favors an adaptive approach over blanket bans. «We can’t pretend AI doesn’t exist, and students will encounter it in their careers. Our job is to teach critical evaluation and responsible use, not to outlaw the technology.» The ethics committee is exploring competency frameworks that distinguish between AI-assisted learning (acceptable, with attribution) and AI substitution for learning (misconduct).

Research integrity presents parallel challenges. A recent controversy involved a PhD candidate who used DALL-E to generate figures for a biology dissertation without disclosing the process. The dissertation committee rejected the thesis pending revised methodology and transparency disclosures. The case highlighted gaps in research ethics policies, which predate generative tools and focus on plagiarism and data fabrication.

Dr. Ruiz is working with the university’s research vice-rectorate to update integrity guidelines, drawing on emerging standards from bodies like the European Network of Research Integrity Offices. Proposed rules would require researchers to document all AI tool usage in methods sections, analogous to disclosing statistical software, and to validate AI-generated content against primary sources.

The governance experiment underway at this Spanish university is far from complete. Dr. Ruiz and her ethics committee are navigating uncharted territory, balancing innovation against risk in an institutional culture built for incrementalism. The outcomes will shape not only their campus but potentially serve as a reference for dozens of universities across Spain and beyond. Whether the model proves replicable or remains an outlier depends on factors beyond any single president’s control: regulatory enforcement rigor, funding availability, and the willingness of faculty and students to accept friction in exchange for accountability. The next academic year will offer critical evidence.