Since the rollout of WCAG 2.2 in late 2023, the publishing industry has made notable strides in implementing accessible practices. But heading into mid-2025, the conversation is evolving. Accessibility is no longer about checklists, it’s about experience, nuance, and scale.
As publishing becomes increasingly AI-assisted, immersive, and multimodal, new challenges are surfacing:
- Fragmentation across platforms and tools
- Inconsistent implementation of accessibility in EPUB 3 and reflowable formats
- Accessibility debt created by AI-generated or auto-tagged content
- Misalignment between legal compliance and actual usability
We’re seeing a wave of challenges emerge that have nothing to do with awareness, and everything to do with execution. From AI content pipelines to EPUB complexities and interactive media that breaks assistive tech, it’s getting harder, not easier, to get accessibility right.
Let’s talk about what really matters for publishers now, and how Integra can help you move forward.
1. Compliance Isn’t the Finish Line. Usability Is.
You can pass an audit and still have an inaccessible product. We’re seeing it all the time:
- Headings look right in code but make no sense to screen readers.
- Alt text is technically present, but completely meaningless.
- Misuse of ARIA labels and headings
- PDFs and EPUBs that pass automated checks but fail user testing
- Navigation is keyboard-friendly, until you open a dropdown.
Take this case: the Table of Contents in an academic EPUB had links applied only to the page numbers, not the headings. So, screen reader users had to hunt with pinpoint accuracy to activate a link that, visually, spanned an entire line.
What we did:
We extended the clickable area to include the full TOC entry, heading and page number. A small tweak, but one with a big payoff.
The result:
- Screen reader users could activate links easily.
- Mobile and tablet navigation became seamless.
- It matched user expectations and reduced friction for everyone.
2. AI-Generated Content Is Creating a New Kind of Accessibility Debt
LLMs are everywhere now. Summaries, image descriptions, learning objectives, more and more of it is machine generated.
As publishers automate alt text, summaries, and image generation, accessibility is often bypassed or faked.
- GPT-generated alt text lacks context
- Auto-captioning misses scientific/educational accuracy
- AI doesn’t catch navigational hierarchy flaws
In one project, an AI tool described an image as:
“An image of a boy.”
But zoom out, and the full scene showed a student with a prosthetic arm working on a STEM project. The detail wasn’t just relevant—it was the point.
What we changed:
We applied human editorial review and rewrote the alt text to:
“A student with a prosthetic arm assembling a battery-powered circuit board.”
Why it matters:
- The instructional value is restored.
- The description respects and includes disability representation.
- It aligns with the learning objective.
AI can move fast, but inclusive storytelling still needs a human touch.
3. EPUB 3 Isn’t the Fix Publishers Were Hoping For
Publishers have embraced EPUB 3 for its flexibility, but real-world usage still reveals gaps.
- TOCs are broken.
- Reflow doesn’t work properly on mobile.
- STEM content is a nightmare for screen readers.
We’ve seen too many EPUBs that technically validate but totally fail the reading experience.
In one scenario, math equations formatted with MathML displayed beautifully on desktop, but completely broke when zoomed or reflowed on mobile. The layout collapsed. The meaning was lost.
How we fixed it:
- Rebuilt equations as inline scalable elements.
- Added descriptive audio-text equivalents.
- Adjusted layout containers to support 200–400% zoom without horizontal scroll.
What changed:
- STEM content became accessible to all users, including those on screen readers or mobile devices.
- No more layout breakage.
- Reflowable content actually, reflowed.
EPUB 3 is a great container, but it’s only as good as the care put into the content.
4. Alt Text at Scale Is Still a Bottleneck
If you’re managing large content backlogs, textbooks, journals, archives, you know the drill:
- Massive backlogs of legacy images
- Low accuracy from internal staff or basic AI tools
- Inconsistent style across alt-text fields
And yet, accessibility guidelines don’t pause for capacity constraints.
One publisher we worked with had thousands of images across multiple disciplines—and a mountain of inconsistent alt text. Some were too vague, others too long, and many just didn’t reflect the actual content.
What we did:
We set up a domain-specific taxonomy for alt text. One set for K–12 education, another for STM, and a third for trade books, each with templates, tone guidelines, and editorial QA.
The impact:
- Alt text aligned with audience expectations.
- Editorial tone became consistent.
- Backlogs started clearing, without sacrificing quality.
Scaling accessibility isn’t about automation alone. It’s about structure, context, and domain fluency.
5. Interactive Content Is Still Largely Inaccessible
We’re seeing a rise in drag-and-drop modules, hotspots, embedded assessments, all great for engagement. But they’re often invisible to assistive tech.
Let’s consider drag-and-drop quizzes, click-to-reveal flashcards, and animation-heavy assessments. They’re engaging, but too often, they’re completely inaccessible.
In one case, a quiz module used only red and green flashes to show right or wrong answers. No alt text. No ARIA roles. No keyboard navigation.
What we changed:
- Introduced text-based feedback with live regions for screen readers.
- Built in keyboard focus cues.
- Paired color with icons and text labels.
The result: A previously invisible experience became fully navigable—without changing the visual design.
Accessibility doesn’t kill interactivity. It enhances it, when it’s part of the design from the start.
6. AI Authoring Tools Are Baking in Problems
Platforms like Adobe Sensei, Canva Docs, and AI-based LMS modules are speeding up content creation. But they’re also embedding issues like:
- Heading abuse (everything’s an H1)
- Auto-generated “help” text that’s vague or unhelpful
- Missing structural tags and metadata
Our approach: We worked with editorial teams to build accessibility-aware prompt libraries, so that AI generates usable structure and inclusive copy from the start.
If AI is writing your content, you need to train it to write accessibly.
7. Tool Fragmentation Is Undermining the Process
Even in the most forward-thinking teams, accessibility lives in silos. Different teams using different tools leads to gaps.
- Editorial uses one CMS.
- QA uses another tool to test output.
- Developers rely on yet another platform.
The result? Issues get lost between handoffs.
Our fix: We built a shared dashboard that tracked accessibility issues by content type, owner, and status, pulling data from all systems into one view.
Accessibility improves when everyone sees the same problem, at the same time.
8. There’s Still No Clear Owner
In many organizations, accessibility is everyone’s job, which often means, it’s no one’s job. Editorial assumes dev has it. Dev assumes UX has it. UX assumes QA is running scans.
Sound familiar?
What works better: We help teams define ownership across functions, create escalation paths, and embed accessibility into content ops, not just product plans.
Accessibility works best when it’s embedded in team culture, not parked in legal documents.
9. General-Purpose Audits Don’t Work for Complex Content
Most accessibility audits aren’t built for publishing. They miss:
- Scholarly references and footnotes
- Glossary structures
- Dynamic content filters
Here’s a real alt text example from a “compliant” PDF:
“please visit and the respective URL”
There’s no context. No clarity. No value.
What it should’ve said:
“Please visit the ABC University Press website”
It’s a tiny change. But for a screen reader user, it’s the difference between orientation and disorientation.
Every element in accessible content should carry meaning, not just pass validation.
10. We’re Still Underserving Neurodiverse Readers
Accessibility often focuses on physical or visual disabilities. But what about:
- Readers with ADHD, dyslexia, or cognitive impairments?
- Learners who need fewer distractions, more spacing, and clearer CTAs?
- Users who struggle with dense or overly complex content?
Our approach: We simplify interfaces, chunk content into scannable sections, and use consistent iconography, all backed by research into neurodiversity and usability.
So What Now?
If you’re in publishing and still treating accessibility like a checklist, you’re behind. The conversation has moved on, and so should your approach.
Here’s what to focus on post-June 2025:
- Audit full user flows, not just templates
- Review AI-generated content for real-world accessibility
- Fix your EPUBs (yes, even the “good enough” ones)
- Give ownership to someone—don’t let it drift
- Start integrating cognitive accessibility into your UX and editorial
- Alignment with international standards like EN 301 549 and EAA
The European Accessibility Act (EAA) goes into enforcement in June 2025. If your organization delivers digital content, platforms, or services within the EU, particularly for the public sector or educational use, you’re now required to comply with EN 301 549, the EU’s harmonized standard for digital accessibility.
This standard builds upon WCAG 2.1 (and now 2.2) but also covers broader ICT contexts, such as e-books, mobile apps, and learning platforms. Many procurement processes in the EU already require EN 301 549 conformance as a baseline.
What this means for publishers:
- Your EPUB 3 files must meet technical and functional criteria under EN 301 549.
- Your platforms, whether LMS, CMS, or digital storefront, must ensure end-to-end accessibility, not just content-level fixes.
- Public-facing and educational services risk exclusion from EU contracts without documented compliance
We’ve been working with publishers for 30 years to know the edge cases. We get the pressure of scale. We understand the friction between good intentions and realistic timelines.
And we’re here to help make accessibility work, not just technically, but practically. Our solutions are:
- WCAG 2.2 & Section 508 compliant
- Built for EPUB 3, LMS, PDF, and HTML5 environments
- Human-reviewed, domain-specific, and audit-backed
- Designed for scale, accuracy, and longevity
Connect with our experts and take a lead on accessibility
