With 95 per cent of undergraduates now using artificial intelligence in some form and 94 per cent using it to help with assessed work, a new article published by the Higher Education Policy Institute (HEPI) contends that universities can no longer afford to treat AI literacy as an enrichment activity. Their argument is that uneven AI capability has become a structural equality-of-opportunity problem — one that existing regulatory frameworks are well placed to address.
The argument draws on HEPI and Kortext’s Student Generative AI Survey 2026, which found that while AI use is near-universal, only 48 per cent of students feel their teaching staff are helping them develop the AI skills they will need in their careers. The survey also found statistically significant differences in the use of text-generation, coding and data-analysis tools across social grades. The implication is that widespread access has not eliminated the digital divide — it has shifted it. The old divide was about whether a student could get online; the new one is about whether they can critically interrogate what an AI system produces, spot a false citation, recognise sycophantic outputs, or judge when not to use the tool at all.
To address this, the article proposes a minimum AI learning entitlement guaranteed to every student by the end of their first term. The entitlement would comprise four elements:
- Equal access to an institutionally supported tool or a verified no-cost alternative
- Discipline-specific teaching on verification, disclosure, data protection and the limits of AI systems used in that field
- Low-stakes practice before students are assessed on their AI use
- Clear assessment rules alongside human review for suspected misuse
The measure of success should not be whether a student can generate a polished output, but whether they can explain why they accepted, rejected or modified what the system gave them.
The piece also sets out a regulatory route for embedding these requirements. In England, access and participation plans already oblige providers to identify risks to equality of opportunity and demonstrate interventions. The Office for Students’ Equality of Opportunity Risk Register explicitly flags unequal knowledge and skills and insufficient academic support as sector-wide concerns. The author argues that uneven AI capability now cuts across both categories, and calls on the Office for Students to name AI capability explicitly in guidance attached to the relevant risk areas. This, the author says, would not create new regulatory bureaucracy but would make visible a mechanism through which existing inequalities are being reproduced.
The core contention is that the next significant divide in higher education will not separate students who use AI from those who do not, but will separate those who can critically evaluate AI outputs from those who cannot — and that universities have both the tools and the obligation to close it.