By Dave Holloway
The rapid evolution of generative AI has fundamentally disrupted traditional higher education frameworks, particularly at the intersection of assistive technology and student support. The sudden, widespread availability of highly competent AI-powered systems has completely upended how we think of accessibility in relation to digital technology and, in some respects, created more barriers to implementation.
Historically, standard software packages solved singular tasks like basic reading or spelling. Today, these same packages boast total document creation engines, rapidly blurring the boundaries between baseline assistance and actual writing. For higher education professionals, navigating this new reality requires a deep understanding of how external funding schemes like the Disabled Student Allowance (DSA) interact with institutional policies.
The Current DSA Process and Institutional Disconnect
To understand the current challenges, we must first look at how the DSA process operates nationwide. The Disabled Students' Allowance is a UK government funding scheme designed to cover key study-related costs arising directly from mental health conditions, long-term illnesses, or specific learning differences, providing tailored software licenses, hardware, and non-medical human help. Students undergo a needs assessment, often before their university place is even confirmed, and Student Finance England (or a regional equivalent) issues an entitlement letter.
At large UK higher education institutions, external bodies fund thousands of DSA students. Crucially, the national procurement model means this funded software is not pre-checked or security-vetted by the university prior to allocation. Furthermore, academic departments are rarely directly notified of a student's DSA funding with tutors typically only seeing an internal Learning Support Plan (LSP), which only lists DSA accommodations if they directly impact the physical classroom environment.
This systemic disconnect introduces significant institutional vulnerabilities across the sector, beginning with data security. There is an extensive, ever-growing list of tools that DSA assessors can potentially recommend to students, ranging from note-taking apps to advanced reading software. Currently, the national framework ensures individual student support needs override institutional security requirements.
Consequently, much of the software recommended and used by students via DSA funding has never been through a formal Data Protection Impact Assessment (DPIA) process at the institutional level. Higher education providers are left in a tricky position where they cannot easily clear or safety-vet these tools, yet they are legally and ethically mandated to accommodate their use to ensure equal access to education.
Feature Creep and Academic Misconduct
Beyond data privacy, the rapid integration of generative AI has introduced a race to include AI functionality across the entire digital ecosystem. Well-established assistive tools are integrating generative AI options overnight without notifying universities, staff, or students. For instance, a student using a tool like Grammarly Pro for mechanical adjustments is now only one click away from generating entire paragraphs of text. This creates massive academic anxiety; students are often completely uncertain where legitimate support ends and academic misconduct begins. It also creates additional anxiety for those students wishing to learn skills manually and become critical thinkers as the opportunities to do so are reducing to the ubiquity of AI functionality.
The problem becomes even more acute in highly sensitive academic environments. In certain professional courses, such as clinical psychology or medical programs where students record real patient sessions, the use of any cloud-based AI tool poses major data privacy and patient consent concerns. If a university restricts these AI-powered DSA tools to protect external data, the institution is still legally mandated to provide alternative adjustments for disabled students. This leaves staff caught between a rock and a hard place, often without clear sector guidance on how to adapt their teaching or what alternative, non-AI assistive technology can fill the gap.
This tool landscape also creates a severe risk of indirect discrimination under the Equality Act 2010. Because declaration rules vary wildly across faculties and institutions, students face an unstable landscape. Forcing a student to declare their DSA-funded AI tools essentially forces them to "out" their disability status to peers and tutors.
Conversely, if a student chooses not to declare a tool and its automated features produce AI-identifiable text, they face an academic misconduct charge. The student is then placed in the deeply unfair position of having to retroactively disclose a disability simply to defend their academic integrity.
Furthermore, the emergence of AI has created extra burdens and exacerbated existing inequalities for disabled students. Across the sector, neurodivergent students are reporting that they are being falsely flagged by AI detectors far more than their neurotypical peers due to structural differences in how they write and organise text (this is a known phenomenon), forcing them to repeatedly defend their baseline integrity.
Subject Failures and Neurodivergent Homogenisation
Student focus groups from the University of Sheffield have shed light on the functional and psychological shortfalls of relying solely on automated digital fixes. From a functional standpoint, a generic "one-size-fits-all" list of AI tools often fails specific disciplines. For example, standard transcription software is frequently regarded as "worthless" by students in Mathematics and Physics because the underlying AI engines cannot reliably recognise formulas, symbols, or units.
In creative and humanities disciplines, AI summarisers often erase crucial subtext and narrative nuance. Psychologically, neurodivergent students have expressed deep concern that AI communication assistants act to "reprogramme" their natural expression to match standard, majority norms. Deployed carelessly, these tools can feel dehumanising, stripping away individual student voices under the guise of making them sound "more professional".
Adding to this instability, the UK government's ongoing consultation on the provision of DSA proposes moving away from funding assistive software as standard, suggesting it should only be funded when free alternatives cannot meet a student's needs. While this threatens to roll back vital support, some experts suggest that redefining DSA might offer an unexpected silver lining for institutions. If universities provide baseline equity by giving everyone access to free or centrally approved tools, the majority of user needs could be met safely and transparently. This would allow institutions to free up time and resources to focus specialised, human-led support on students with the most complex needs.
Moving Forwards
To navigate these overlapping crises, institutional change must begin in the classroom. Academic and support staff should actively audit the technical software used in their modules for hidden AI feature creep and explicitly clarify to students what functionalities are permitted within specific assignments.
Students must also be empowered with clear boundaries to secure their academic standing. Institutions should guide students to isolate mechanical support (like spelling and grammar adjustments) from actual content generation when using advanced tools. Students should also be encouraged to proactively share their external DSA entitlement letters with internal disability support services to streamline their accommodations safely. Guidance for disabled and neurodivergent students should be mapped directly to pre-existing DSA categories to ease any transitions, and communicated on a practical, task-by-task basis.
While institutionally provided AI platforms can be leveraged by individuals and teams to replicate basic functions, they cannot fully replace dedicated, specialised tools. Moving forward, higher education institutions must involve students in policy co-creation, ensuring future AI and accessibility guidelines are built directly around lived neurodivergent experiences rather than automated, one-size-fits-all fixes.