
As of September 2026: State and local governments must meet WCAG 2.1 Level AA under ADA Title II, on a timeline the DOJ pushed to April 26, 2027 and April 26, 2028. Employee training itself is governed separately, under ADA Title I, which requires employers with 15 or more staff to give employees with disabilities equal access to training, without specifying a technical web standard. Public-facing digital services from private businesses fall under ADA Title III, which has no fixed deadline but is increasingly measured against WCAG 2.1 AA in court. Federal agencies are directly subject to Section 508 for covered ICT, and vendors supplying ICT to federal agencies are commonly required through procurement to demonstrate Section 508 conformance. The current legal baseline either way is WCAG 2.0 Level A and AA, not 2.1 or 2.2.
Here's what most compliance checklists published before April 2026 got wrong: the federal deadline they warned you about already moved. The DOJ pushed its ADA Title II deadline for large public entities from April 24, 2026 to April 26, 2027, and smaller entities now have until April 26, 2028. If your only accessibility trigger was that date, you technically have more time.
That extension doesn't touch the two frameworks most likely to matter for a corporate training video: Title I, which sets your obligations to your own staff, and Title III, which covers anything you show customers or the general public. Neither has ever had a fixed compliance date. Title III specifically is enforced through litigation, and that litigation didn't slow down while Title II's deadline moved. Federal courts saw 3,117 website accessibility lawsuits in 2025, a 27% jump from 2024, according to Seyfarth Shaw's tracking of ADA Title III filings. Those are general website accessibility cases, not corporate-training-specific ones, but courts increasingly point to WCAG 2.1 Level AA as the practical benchmark for "accessible" across this kind of claim, deadline or no deadline.
This piece covers which legal framework actually applies to your training video, what WCAG 2.1 AA requires in practice, what's different about interactive video specifically, and a practical checklist for auditing what you already have. It closes with a direct account of what Clixie AI supports today and what still needs confirming, because overclaiming accessibility to a compliance audience is a specific way to get tested.
Key Takeaways

Five things, in plain terms: synchronized closed captions, audio description for prerecorded video, accessible alternatives for any meaningful visual information the audio doesn't cover, a keyboard-operable player with visible focus, and no content that flashes more than three times per second unless it falls within WCAG's defined flash-size and color thresholds. A plain transcript is good practice everywhere, but it is not itself the Level AA requirement, which is a more precise point than most checklists make.
WCAG organizes its rules under four principles, commonly shortened to POUR: Perceivable, Operable, Understandable, and Robust. The video-specific success criteria sit mostly under Perceivable, and they overlap in a way that trips people up. W3C's own Understanding document for SC 1.2.5 lays it out precisely:
LevelRequirement for prerecorded videoA (SC 1.2.3)Either an audio description or a full text alternative for the visual content, author's choiceAA (SC 1.2.5)Audio description specifically. If you satisfied Level A with a text alternative instead of audio description, AA adds audio description as a separate requirementAAA (SC 1.2.8)A full text alternative (extended text description), on top of whatever satisfied A and AA
The detail worth sitting with: an ordinary transcript, the kind that just captures dialogue, generally doesn't satisfy any of these on its own. The "text alternative" in SC 1.2.3 has to describe visual information too, not just repeat the audio, and the AA requirement (1.2.5) specifically wants audio description, not a document to read separately. Most legal standards, including the DOJ's Title II rule, target Level AA. That's the bar this pillar uses throughout.
Important exception: W3C states that additional audio description isn't necessary when all important visual information is already communicated in the video's existing audio. This matters more for training video than for entertainment media, because instructional narration can often be written from the start to describe the meaningful action on screen, "as you can see, the technician now checks the pressure gauge on the left," which satisfies the intent of SC 1.2.5 without a separate description track.
One myth worth killing early: automatically generated captions should not be assumed to be accurate enough without review. WCAG requires captions to convey the spoken dialogue and relevant non-speech audio needed to understand the content, things like a door slam or an alarm, but it does not prescribe a numeric accuracy percentage. Services such as 3Play Media advertise 99% accuracy for human-reviewed captioning, which illustrates the quality professional captioning workflows target, not a formal WCAG threshold. Most auto-generated captions fall short of that vendor benchmark without a human review pass, a point the next section covers in more detail.
Does the ADA actually require your training video to be accessible? It depends which framework applies, and corporate training video can touch more than one at once: internal employee training sits under one part of the law, and anything you show to customers, prospects, or the public sits under another.
SituationGoverning frameworkTechnical benchmarkDeadline or statusEmployee training and internal L&DADA Title INo WCAG version specified; reasonable-accommodation contextNo fixed deadline; enforced via EEOC complaintsState and local government digital contentADA Title IIWCAG 2.1 Level AAApril 26, 2027 (large entities) / April 26, 2028 (smaller entities), after the 2026 extensionPublic-facing private-sector content (marketing, sales, ecommerce training)ADA Title IIINo federally codified WCAG version; courts commonly reference 2.1 AANo fixed compliance deadline; enforced through DOJ action and private litigationFederal agency ICT and vendor contentSection 508WCAG 2.0 Level A and AA, incorporated by referenceExisting requirement, no recent change
Title I governs your obligations to your own employees, and it doesn't specify a web accessibility standard. ADA.gov's guide to the ADA is explicit that Title I "requires employers with 15 or more employees to provide qualified individuals with disabilities an equal opportunity to benefit from the full range of employment-related opportunities... including... training." That's a reasonable-accommodation obligation enforced by the EEOC, not a WCAG conformance mandate. In practice, following WCAG 2.1 AA is still the most defensible way to meet that obligation for video-based training. It's the closest thing to an accepted technical definition of "accessible" that exists. Just know this is the framework actually governing internal training, not Title III.
Title II covers state and local government entities, and it does have deadlines, but they just moved. The DOJ's original 2024 rule set April 24, 2026 as the compliance date for public entities serving 50,000 or more people, with April 26, 2027 for smaller entities and special districts. Then, on April 20, 2026, four days before that first deadline hit, the DOJ issued an Interim Final Rule extending both dates by roughly a year. Large entities now have until April 26, 2027. Smaller ones have until April 26, 2028. The technical standard itself, WCAG 2.1 Level AA, didn't change. Only the enforcement clock did.
Title III covers public-facing private-sector content, and it has no fixed deadline at all. If your training video is customer-facing, part of a course you sell, or embedded in a public marketing or ecommerce experience, this is the framework that applies, not Title I. Title III itself never specified WCAG as the technical standard. Courts have converged on WCAG 2.1 AA anyway, using it as the practical yardstick for what "effective communication" and "equal access" mean in digital accessibility cases. With no compliance date to work toward or wait out, private litigation has become a major enforcement mechanism for digital accessibility claims, alongside the DOJ's own authority to bring civil actions directly. Either route is active right now.
Section 508 directly governs federal agencies, and its legal bar is older than most people assume. U.S. Access Board's Revised 508 Standards incorporate WCAG 2.0 Level A and AA by reference, not 2.1 or 2.2, for ICT that agencies develop, procure, maintain, or use. A private company doesn't "fall under" Section 508 the same way an agency does; vendors supplying ICT to federal agencies are commonly required through procurement, not the statute directly, to demonstrate Section 508 conformance, typically via a VPAT. If Clixie or your organization sells into or contracts with a federal agency, targeting a newer WCAG version is good practice and increasingly expected in those procurement conversations, but it isn't yet the codified legal minimum.
The DOJ was explicit about why it extended the Title II deadline: it had overestimated how far automated remediation technology had come, and many public entities didn't have the staff or budget to hit the original dates. That reasoning doesn't touch Title I or Title III at all. Neither framework was on that clock, and the National Federation of the Blind sued DOJ and HHS over the extension in May 2026, arguing the agencies bypassed public comment and understated the harm to disabled users. Whichever side of that fight you land on, the practical takeaway is the same: the deadline that moved was never the one governing employee training or public-facing content to begin with.
.avif)
Because the deadline that moved only applies to Title II, and these lawsuits are almost entirely Title III cases against private businesses. Federal website accessibility filings climbed to 3,117 in 2025, up from 2,452 in 2024, a 27% increase, according to Seyfarth Shaw's annual tracking. Those cases made up 36% of all federal ADA Title III lawsuits filed that year. Industry trackers that also count state-court filings put the 2025 combined total above 5,000.
Worth being precise about what this statistic actually shows: it's general website and digital-service accessibility litigation, concentrated heavily in ecommerce, not lawsuits specifically about employee training video. There's no public data showing how many of these cases involve internal training content, and this article won't imply otherwise. What the trend does establish is the legal climate. Courts hearing Title III claims increasingly treat WCAG 2.1 AA as the working definition of "accessible." That climate applies to any public-facing digital content a business publishes, training included, whenever that training is customer-facing or otherwise public.
For internal employee training specifically, the more direct exposure runs through Title I and the EEOC's reasonable-accommodation framework, not through this website-lawsuit dataset. Either way, there's a completion-tracking angle worth remembering. If a mandatory course genuinely can't be completed by an employee using a keyboard or screen reader, that's not just an accessibility gap. It's also a gap in the exact completion record compliance teams rely on to prove training happened.
Because branching paths, quizzes, and hotspots add failure points that a plain video doesn't have. Captioning solves the audio problem. It does nothing for a decision point that's communicated only through a color-coded arrow or a click target a mouse-only interaction requires.
Picture a branching compliance scenario where the learner chooses how to respond to a vendor gift offer. If that choice is presented as two colored buttons with no text label and no keyboard focus indicator, a screen reader user has no way to know the options exist, and a keyboard-only user has no way to select one. What happens next? The video plays captions correctly right up to that point, and still fails the learner completely at the one moment the training actually tested something.
The same problem shows up in quizzes and hotspots. A hotspot that only appears on hover has no keyboard equivalent at all. A multiple-choice quiz where the correct answer is marked only by a green highlight, with no text confirmation, tells a colorblind or screen-reader-dependent learner nothing. None of these are exotic edge cases. They're standard interactive video features, built the same way across most authoring tools, and none of them were designed with a non-mouse user in mind.
.avif)
Cognitive accessibility is part of this too, and it overlaps with good learning design generally. Short sentences, plain instructions, and consistent navigation patterns aren't just accessibility requirements, they're the same principles that make a branching scenario easier for anyone to follow under time pressure. Accessible design and well-designed practice questions solve adjacent problems with the same fix.
Start with your highest-traffic mandatory courses, not your entire library. Run an automated scan first to catch the easy issues, then test manually with a keyboard only and, ideally, a screen reader.
Automated tools matter, but how much they catch depends entirely on how you count. By one measure, counting how many WCAG 2.1 AA success criteria a scanner can test at all, Deque found its own axe-core engine could partially or fully test only 15 to 16 of 50 criteria, roughly 30%. Deque itself argues that measure understates real-world value. In a separate analysis of over 13,000 first-audit pages, it found automated scans caught 57% of actual issues by volume. The gap comes down to frequency: a handful of common, easily detected problems, color contrast being the classic case, make up a disproportionate share of what's actually wrong on real pages (Deque's own coverage study). A separate UK Government Digital Service test, which deliberately planted 142 known barriers on a single page and ran 13 different checkers against it, found the best-performing tool caught 40% of them. All three numbers are legitimate and they're measuring different things: success criteria coverage, issue volume, and detection of known planted barriers. What they agree on is the conclusion: a clean automated scan is not proof of an accessible course. Things like whether alt text is actually meaningful or whether a focus order makes logical sense still require a human.
.avif)
A practical audit sequence:
.avif)
No, not on their own. WCAG expects captions to be accurate, synchronized, and to include speaker identification and relevant non-speech sound cues. Most auto-caption tools produce a reasonable first draft of the dialogue and stop there: no speaker labels, no sound descriptions, and accuracy that falls short of the 99% benchmark captioning vendors like 3Play Media guarantee on human-reviewed work.
The practical implication is a workflow, not a tool choice. Auto-generated captions are a legitimate starting point. Treating them as the finished, compliance-grade artifact is where teams get exposed. A short human review and correction pass, checking names, technical terms, and speaker changes, is what turns a fast first draft into something that holds up.
Organized under the four WCAG principles:
Perceivable
Operable
Understandable
Robust
Which of these is worth fixing first? Captions and transcripts are almost always the fastest wins, since they're a content fix rather than a redesign. Keyboard operability on branching choices takes longer because it touches the interaction itself, not just the media around it. This is the section worth turning into a standalone checklist your team can run against any existing course before it becomes this pillar's downloadable asset.
Here's what's confirmed, mapped directly to the requirements above, and what still needs a straight answer from product before it's claimed anywhere else.
WCAG requirementWhat it needsHow Clixie helpsSynchronized captionsAccurate, timed captions with speaker identificationGenerates captions automatically from the video's transcript, with a player-level toggle rather than captions burned permanently into the videoMulti-language accessCaptions or subtitles in the learner's languageSupports subtitles in 120+ languages, generated automatically from the source videoDialogue transcriptA text record of spoken contentGenerated alongside captions from the same pass. This is good practice but is not itself the WCAG AA audio-description requirement (SC 1.2.5), which needs visual content described, not just dialogue transcribedAudit-ready compliance recordsProof that training was completedSCORM export includes completion tracking, so accessibility remediation and its usage record live in the same audit trail compliance teams already report from
Three things are not yet confirmed, and this article won't claim them until they are. Is the Clixie player itself fully keyboard-operable end to end, including branching choice points? Does it expose ARIA roles that screen readers interpret correctly? Is audio description natively supported, or does it require a manual workaround today? Each needs a straight answer before it appears anywhere else in Clixie's marketing. If Clixie has a third-party accessibility audit or a Voluntary Product Accessibility Template on file, that's worth linking here directly, since it's the single most credible thing a compliance buyer could see in this section.
This is deliberate. Compliance and accessibility buyers are exactly the audience most likely to test a claim like "fully accessible" themselves, and getting caught overstating it costs more credibility than being precise costs momentum.
Yes, for employers with 15 or more employees. Title I requires equal access to training as part of employment, enforced by the EEOC through reasonable-accommodation obligations. It doesn't specify WCAG as a technical standard the way Title II does, but WCAG 2.1 AA is still the most defensible way to demonstrate that a training video is accessible to an employee who needs it.
It depends which framework applies. State and local government content falls under ADA Title II, which now targets WCAG 2.1 AA by April 26, 2027 or April 26, 2028, after the DOJ's 2026 extension. Employee training falls under Title I, which doesn't name a technical standard. Public-facing private content falls under Title III, enforced through litigation that increasingly cites WCAG 2.1 AA. Federal agencies are directly subject to Section 508; vendors reach it through procurement requirements when they sell ICT to those agencies.
No. The current Revised 508 Standards incorporate WCAG 2.0 Level A and AA by reference, not 2.1 or 2.2. Organizations may still choose, or be contractually required by a specific agency, to target a newer WCAG version as a procurement expectation, but that's separate from the codified legal minimum.
Not on their own. WCAG expects captions to be accurate, synchronized, and to include speaker identification and sound cues. Auto-generated captions are a reasonable first draft but typically need a human correction pass before they meet the accuracy bar compliance-grade captioning requires.
At Level A (SC 1.2.3), you can choose either audio description or a full text alternative for a prerecorded video's visual content. At Level AA (SC 1.2.5), audio description specifically is required, an ordinary transcript doesn't satisfy it on its own. Most legal standards, including the DOJ's Title II rule, target Level AA.
Only if they sell to or contract with federal agencies. Otherwise, ADA Title I (employee training) or Title III (public-facing content) are the relevant frameworks for private businesses, and neither specifies a WCAG version in statute, though courts commonly reference WCAG 2.1 AA for Title III claims.
Every choice point needs a text label, not just a color or icon, and needs to be reachable and selectable with a keyboard alone. Screen reader users need the same information sighted, mouse-using learners get: what the choice is and what happens when they select it.
Audit your highest-risk courses first: anything tied to a legal requirement or with the largest completion volume. Fix captions and transcripts immediately since it's the fastest win available. Then treat keyboard and screen reader testing as a standing QA step for every new course, not a one-time cleanup project.
The deadline that made headlines moved. The one that matters to most corporate training teams never had a date to begin with.
Start with one course: run an audit, fix captions and transcripts first, and test the branching points with a keyboard alone.
Ready to see where your existing training video stands? Start a free trial with Clixie AI and run your first course through it.
Jancarlo Nesto is Co-Founder and COO of Clixie AI, an interactive video platform used to build training, onboarding, compliance, sales, and educational video experiences. He has spent more than 20 years in software product strategy, go-to-market, and leading distributed teams across Latin America, North America, and Europe, with prior leadership roles at Eidosmedia, Atex Group, and Unisys. He holds a degree in electrical engineering and an MBA in project management. This article combines Clixie AI's product knowledge of interactive training video with primary accessibility guidance from the U.S. Department of Justice, W3C, and the U.S. Access Board, cited throughout.
This article was reviewed against accessibility standards and government guidance available as of September 21, 2026. Primary sources were prioritized, including ADA.gov, the Federal Register, W3C's Web Content Accessibility Guidelines, and the U.S. Access Board's Section 508 standards. Legal requirements, technical accessibility standards, recommended practices, and Clixie's own product capabilities are treated as separate categories throughout, not blended into one another. Statistics from accessibility vendors, law firms, and industry research are attributed to their original sources rather than presented as regulatory requirements.
Clixie AI develops interactive video software used to add captions, transcripts, branching scenarios, and SCORM tracking to existing video content. Some subjects in this article, including captions, transcripts, branching scenarios, and SCORM export, overlap directly with Clixie's product. Product capabilities are stated only where they're confirmed on Clixie's public product pages as of publication; capabilities that require product or legal confirmation are identified as unverified rather than asserted. Using Clixie does not by itself make a training program compliant with the ADA, Section 508, or WCAG; compliance depends on the underlying content, interaction design, and an organization's specific obligations. This article is not a substitute for a legal accessibility review of your own training content.
Articles involving accessibility, compliance, or regulatory requirements are reviewed when material standards or government guidance change. Publication and last-updated dates are provided so readers can judge the currency of the information.
Last reviewed: September 21, 2026
Legal notice: This article is for educational purposes and does not constitute legal advice. Organizations should consult qualified accessibility and legal professionals when evaluating their specific compliance obligations.