Accessibility is Infrastructure, Not a Final Check

Modern publishers are rethinking where accessibility belongs in the production process

For decades, accessibility in publishing has often been treated as the final stage of production. Manuscripts are edited, designed and converted into digital formats, and only then is accessibility considered. At best, this approach introduces additional remediation work. At worst, it produces content that is fundamentally difficult to adapt, delaying delivery and increasing costs.

That workflow is no longer viable.

Educational publishers now operate in an environment defined by expanding accessibility regulation, diverse learning needs, and rapidly accelerating content production cycles. Recent regulatory changes are making the stakes concrete. In the US, ADA Title II now requires public sector organisations, including public universities and institutions, to ensure their digital content is accessible, with compliance deadlines set for April 2026, covering websites, platforms, and educational materials such as PDFs and digital textbooks. In the EU, the European Accessibility Act came into effect on 28 June 2025, with its scope covering services across education and digital content, and applying to any organisation conducting business in the EU, not just those based there.

For publishers, the combined effect of these regulations is direct. Accessibility is no longer just an internal production consideration; it is increasingly a condition of market access. When accessibility is treated as an integrated design principle rather than a post-production check, publishers gain not only compliance but operational efficiency. When it is treated as an afterthought, the opposite occurs: remediation, risk and delay.

Accessibility failures are often workflow failures

Accessibility problems rarely arise because publishers lack good intentions. They arise because of how publishing workflows are structured.

Traditional content pipelines are highly sequential. Authors produce manuscripts. Editorial teams refine them. Production teams convert them into formats suitable for print or digital delivery. Accessibility specialists then attempt to retrofit compliance requirements onto the finished output.

This model introduces structural friction at every stage.

Images are created without alternative text guidance
Tables are structured visually rather than semantically
Heading hierarchies become inconsistent through editing
Metadata required for accessibility is incomplete or missing

By the time accessibility teams review the material, correcting these issues often requires reworking content that has already passed through multiple stages of production.

The World Wide Web Consortium stresses that accessibility should be embedded within design and development workflows rather than treated as a retrofit exercise.

Experience consistently shows that fixing accessibility after publication is far more expensive than designing for it from the start.

In educational publishing, where content is repeatedly revised, repurposed, and distributed across multiple digital formats, the operational cost of late-stage remediation compounds quickly.

AI has changed what is operationally possible

The rapid development of AI-assisted publishing workflows has fundamentally changed the economics of accessibility.

Tasks that once required significant manual effort can now be completed at scale:

Generating structured metadata
Creating draft alternative text
Checking heading hierarchy and semantic structure
Flagging accessibility violations across large content collections

Publishers are increasingly experimenting with AI-assisted workflows to automate metadata, editorial processes, and production tasks across the content lifecycle, as documented by both the Book Industry Study Group and Springer Nature's own published reporting on its editorial operations.

However, the real opportunity lies not in automation but in timing.

AI systems can analyse and enrich content as it moves through the production pipeline rather than at the end. Accessibility checks can occur simultaneously with editorial work. Metadata can be generated alongside content creation. Structural validation can run continuously as assets evolve.

This transforms accessibility from a corrective exercise into an operational capability.

If AI can analyse thousands of pages in minutes, there is little justification for accessibility problems to remain undetected until the final production stage.

Accessibility still requires human judgement

Despite these advances, accessibility cannot be fully automated. AI systems can generate elements such as alt text, structural tags and metadata, but human review remains essential.

Two risks stand out. The first is hallucination. AI can produce descriptions that sound convincing but are factually wrong. In educational publishing, that is not acceptable.

The second is contextual understanding. Accessibility is not just technical; it is about how content supports learning. A science diagram needs a very different description from a history image. Only someone with subject knowledge can judge whether it actually helps the learner.

For this reason, human oversight must remain central to AI-enabled publishing workflows.

The most effective approach is a human-in-the-loop model in which AI performs large-scale analysis and draft generation, while expert reviewers establish acceptance criteria and validate the output. This combines the speed of automation with the judgement that educational content demands.

From Compliance to Core Capability

Accessibility is not just about compliance. It is directly connected to educational outcomes. When learning materials cannot be accessed by all students, publishers fail to meet the fundamental purpose of educational publishing.

The growth of AI in publishing workflows only raises the bar. Faster, more efficient production is now possible at scale, which means there is no credible justification for accessibility to remain a slow, manual afterthought.

Accessibility cannot be bolted onto a process that was not designed for it. It must be built into the infrastructure that governs how content is created, structured and reviewed: systems capable of analysing content continuously, applying accessibility rules as material evolves, and supporting human experts in validating the output. In practice, this means moving away from fragmented tools and manual remediation toward integrated content intelligence that understands publishing frameworks, accessibility standards and editorial workflows.

When accessibility becomes an operational capability rather than a final inspection stage, publishers gain more than compliance. They gain faster production, lower remediation costs and confidence that their content can reach every learner it is intended to serve.

In the AI era, accessibility should not slow publishing down. Done properly, it becomes part of what makes publishing scale.

About the Author

Gillian Malone-Johnstone is Head of Customer Success at Syllabyte, where she helps publishers leverage AI-driven solutions for standards alignment, metadata optimisation, and accessibility. With over a decade bridging AI-powered adaptive learning systems and real-world publisher workflows, Gillian has spent her career proving that the best educational technology amplifies human expertise rather than replacing it. Connect with her on LinkedIn.

References

European Commission (2023) European Accessibility Act (Directive (EU) 2019/882). Available at: https://ec.europa.eu/ (Accessed: 9 March 2026).

Horn, B. (2025) AI Use Across the North American Book Industry: Survey Results Presentation. Book Industry Study Group (BISG). Available at: https://knowledgecenter.bisg.org/25acbpo/ (Accessed: 7 April 2026).

W3C (2023) Web Content Accessibility Guidelines (WCAG) 2.2 Overview. World Wide Web Consortium. Available at: https://www.w3.org/WAI/standards-guidelines/wcag/ (Accessed: 9 March 2026).