This is general information about how video accessibility law and standards work, not legal advice. Whether your specific videos, website, or organization has a compliance obligation — and what satisfies it — is a legal determination that depends on facts a guide can't know: your jurisdiction, your entity type, who your content reaches, and case law that keeps moving. If you've received a demand letter, are facing a lawsuit, or need a defensible compliance posture, talk to a lawyer or an accessibility consultant who can look at your actual situation. Everything below is context to make that conversation more useful, not a substitute for it.
WCAG: the technical standard everything else points back to
The Web Content Accessibility Guidelines (WCAG), maintained by the W3C, are the technical backbone that most legal frameworks defer to when they need to define "accessible" in concrete terms. WCAG organizes requirements as testable "success criteria," each assigned a conformance level — A (minimum), AA (the level almost every legal and regulatory framework actually cites), or AAA (aspirational, rarely mandated). Three criteria matter directly for video:
| Criterion | Level | What it actually requires | |---|---|---| | 1.2.2 Captions (Prerecorded) | A | Prerecorded video with synchronized audio must have captions — a synchronized text alternative that includes dialogue and relevant non-speech information (sound effects, speaker changes), not just a rough transcript. This is the baseline requirement almost every video compliance conversation starts and ends with. | | 1.2.4 Captions (Live) | AA | The same idea, extended to live audio content in synchronized media — live streams, webinars, live broadcasts. Real-time captioning, not a caption file added afterward. | | 1.2.5 Audio Description (Prerecorded) | AA | Prerecorded video needs audio description for visual information that isn't conveyed by the existing audio track — what's happening on screen that a blind or low-vision listener would otherwise miss. This is a separate accessibility need from captions and gets confused with it constantly; captions serve viewers who can't hear, audio description serves viewers who can't see. |
The practical read: 1.2.2 is Level A, meaning it's about as close to a floor as WCAG gets — if a site claims WCAG 2.1 AA conformance (which is what nearly every regulation and settlement actually cites), captions on prerecorded video are non-negotiable. 1.2.5 is the criterion organizations most often miss entirely, because captions get treated as "the accessibility box" for video and audio description doesn't come up until someone asks about it directly, or a complaint names it specifically.
The ADA: no caption mandate in the text, real exposure anyway
The Americans with Disabilities Act (US, 1990) is where a lot of confusion starts, because the statute itself is old enough that it never mentions captions, websites, or streaming video — none of that existed in its current form when it was written. Title III of the ADA prohibits disability discrimination in "places of public accommodation" — historically read as physical spaces: stores, restaurants, hotels, theaters.
What's changed is how courts and the Department of Justice have applied that language to the web. Over roughly the last decade, DOJ guidance and a growing body of case law have extended Title III's reach to websites and, by extension, the video content on them — the reasoning being that a business's website functions as an extension of the public accommodation it operates, and inaccessible video is a barrier to that access the same way a step without a ramp would be. The DOJ has formally reinforced this position for public-sector entities, and website-accessibility litigation citing the ADA against private businesses has been one of the more consistent categories of disability-rights litigation in the US for several years running — enough that it functions as a real, recurring litigation category, not a hypothetical.
The result is a genuinely awkward legal position for anyone running a commercial or public-facing site: there's no line in the ADA that says "caption your videos," but there's a substantial and growing body of enforcement and litigation treating uncaptioned video as exactly the kind of access barrier Title III is meant to prevent. Courts don't apply this uniformly — outcomes vary by circuit, by the specifics of the business, and by how the plaintiff frames the claim — which is precisely why "does the ADA require captions" doesn't have a clean yes/no answer and why this is a question for someone who tracks the current case law in your jurisdiction, not a general guide.
Section 508 and EN 301 549: the standards that actually spell it out
Two frameworks skip the ambiguity above because they're written as explicit technical standards rather than broad anti-discrimination statutes.
Section 508 (an amendment to the US Rehabilitation Act) requires US federal agencies, and companies that sell technology or services to federal agencies, to make their electronic and information technology accessible. Since 2018, Section 508's technical standard is WCAG 2.0 Level AA — so a federal contractor's video content has a direct, explicit obligation to meet the WCAG caption and audio-description criteria above, with no ambiguity about whether it applies.
EN 301 549 is the EU/UK equivalent — a harmonized technical standard for the accessibility of ICT products and services, and it's also built on WCAG conformance (2.1 AA) as its core content requirement. It's the technical standard referenced by the EU's public-sector accessibility directive and increasingly by the European Accessibility Act for private-sector digital services and products sold into the EU market.
The pattern across both: where a technical standard exists, it converges on the same target — WCAG 2.1 AA, which means real captions on prerecorded video and audio description where visual information isn't otherwise conveyed. The difference from the ADA isn't the destination, it's that Section 508 and EN 301 549 tell you plainly, in writing, that you have to get there.
Who this actually applies to in practice
Stacking the frameworks above, obligation runs on a rough gradient rather than a single bright line:
- Public entities — government agencies, public schools, municipal services — have the clearest and most direct obligations under both US and EU frameworks, reinforced by explicit DOJ rulemaking for state and local government sites.
- Federal contractors and agencies (US) have an explicit, written Section 508/WCAG 2.1 AA obligation with no interpretive gap.
- Higher education — public and increasingly private institutions — has been a consistent target of ADA-based accessibility complaints and settlements over course video, lecture capture, and online learning platforms.
- Large private businesses with a public-facing web presence carry real exposure under the ADA's Title III interpretation described above, even with no video-specific statutory line to point to. Size and reach matter in practice — a business the DOJ or a plaintiff's firm would recognize as effectively a public accommodation is a more likely target than a two-person shop, but "we're small" isn't a legal shield, just a lower-probability profile.
- Anyone who's received a demand letter or been named in a suit is, by definition, already past the theoretical-risk stage and needs a lawyer, not a guide like this one.
Outside those categories, obligation gets genuinely murkier — which is exactly the point at which "talk to a consultant" stops being boilerplate advice and becomes the actually correct next step.
The practical compliance path for an existing video library
Say the legal question is settled — your organization has decided it needs real, accurate captions on its video library, whether because a consultant said so, a complaint arrived, or leadership just decided not to wait for one. The next question is purely practical: how do you actually get there without commissioning full professional captioning on every video from scratch, which typically runs somewhere in the low single-digit dollars per minute and adds up fast across a real library.
For video that already has auto-generated captions — which by 2026 is most video on YouTube, and increasingly on other platforms — the fastest and cheapest real path is usually extraction and proofreading, not fresh transcription. The auto-caption engine already did the first-draft work; what's missing is a human accuracy pass and, per 1.2.2 above, the non-speech information (speaker identification, meaningful sound cues) that auto-captions typically don't include at all. Pulling the existing caption track out as an editable file and correcting it against WCAG's actual requirements is a fundamentally different — and much smaller — job than transcribing a video with nothing to start from. The manual vs. automatic transcription time guide breaks down why extraction, generation, and manual transcription are three different operations with wildly different time costs, which is the same math that makes proofread-the-draft the cheaper compliance path in most cases.
SubExtract's YouTube caption extractor pulls a video's existing caption or transcript track — whichever the video actually has, auto-generated or human-uploaded — as plain text or a timestamped .srt file, which is the format most caption editors and players expect. It doesn't generate new captions or fix accuracy on its own; it gets you the starting file to proofread against, which for a library that already has some captions is most of the time savings. The free video accessibility captions how-to walks through that extraction-and-proofread workflow step by step, and the accessibility teams guide covers the same problem from the angle of someone managing this across a whole content library rather than one video at a time.
Why unproofed auto-captions aren't a safe compliance answer on their own
The honest caveat, and the one easiest to skip past: auto-generated captions, used as-is with no review pass, are not a defensible compliance position. Auto-caption accuracy runs roughly 90-95% on clear audio and drops meaningfully with accents, technical or industry-specific vocabulary, music beds, or lower audio quality — and that's before accounting for what auto-captions structurally omit, like speaker identification and meaningful sound cues, both of which WCAG's caption criteria expect where they carry information the viewer needs.
A 90-95% accurate transcript sounds close to done. In caption terms it means roughly one error every one or two sentences — garbled names, dropped negations that flip a sentence's meaning, technical terms mangled into something unrelated. For casual viewing that's often tolerable. For a caption track standing in as an organization's accessibility compliance evidence, it isn't — an inaccurate or incomplete caption is exactly the kind of gap that shows up in a complaint or a demand letter, because "we had captions" and "we had accurate captions" are different claims, and only the second one actually satisfies 1.2.2 in any meaningful sense. Extraction gets you the raw material fast and cheap. The proofread pass — checking against the actual audio, adding sound cues and speaker IDs where they're missing, fixing errors — is the part that turns a draft into something that would hold up if it were ever actually reviewed. Skipping it defeats the point of doing the work at all.
One more time, plainly: none of the above is legal advice, and it isn't a substitute for one. Standards, case law, and enforcement priorities move — this guide reflects a snapshot, and the specific determination of whether your organization has an obligation, and whether your captions satisfy it, needs a lawyer or accessibility consultant who can evaluate your actual content, audience, and jurisdiction. Treat this as background for that conversation, not as the conversation's conclusion.
Frequently asked questions
Does the ADA legally require closed captions on video? The ADA's text doesn't mention captions or video directly — it predates both in their current form. But DOJ guidance and a substantial line of court decisions have extended Title III's public-accommodation requirements to websites and their video content, meaning organizations serving the public online carry real exposure even without an explicit statutory caption mandate. This is an area of active litigation and evolving case law, not a settled bright line — get a lawyer's read for your specific situation.
What's the actual difference between WCAG, ADA, Section 508, and EN 301 549? WCAG is the technical standard — the specific, testable rules (like 1.2.2 and 1.2.5) that define what "accessible" means in practice. The ADA is a broad US anti-discrimination law that courts have applied to the web without ever spelling out a technical standard of its own. Section 508 (US federal) and EN 301 549 (EU) are the frameworks that close that gap — both explicitly adopt WCAG 2.1 AA as their technical requirement, so there's no ambiguity about what "accessible" means once one of them applies to you.
Are auto-generated YouTube captions good enough for ADA or WCAG compliance? Not on their own. They're a reasonable starting draft — typically 90-95% accurate on clear audio, worse with accents, jargon, or poor audio — but they're missing speaker identification and meaningful sound cues that WCAG's caption criteria expect, and the accuracy gap alone is enough that most accessibility consultants wouldn't call raw auto-captions defensible. Extracting and proofreading them is a legitimate and much cheaper path than fresh professional captioning; using them completely unreviewed is not a safe compliance answer.
Does this apply to a small business or independent creator? Formally, obligation tends to track public-facing reach and entity type — public agencies and federal contractors have explicit, unambiguous requirements; large commercial businesses carry real ADA-based litigation exposure; a small creator's legal exposure is generally lower but isn't zero, and "small" isn't a defined legal exemption anywhere in these frameworks. Practically, captioning is also just good practice regardless of legal obligation — most social video is watched with sound off. Whether a specific small operation has an actual legal duty is exactly the kind of question a consultant, not a guide, should answer.
What about audio description — is that actually required too? WCAG 2.1's 1.2.5 (Level AA) requires audio description for prerecorded video where visual-only information isn't already conveyed by the existing audio — narration that describes what's happening on screen for viewers who can't see it. It's a separate criterion from captions and gets missed constantly because captions absorb most of the attention in accessibility conversations. Any organization working toward WCAG 2.1 AA conformance — which is what Section 508 and EN 301 549 both point to — has an audio description obligation alongside the caption one, not instead of it.
Next steps
Start with what your video library already has: most video published in the last several years already carries some form of auto-generated captions, and extracting that existing track with the YouTube caption extractor is a faster, cheaper starting point than commissioning fresh professional captioning across an entire library. The free video accessibility captions how-to walks through that extraction-and-proofread workflow in detail, and the accessibility teams guide covers running this across a full content library rather than one video at a time. For the difference between what auto-captions actually contain versus what real closed captions require, see the closed captions vs subtitles guide — and for the time math behind why proofreading a draft beats transcribing from scratch, the manual vs. automatic transcription time guide has the numbers.
None of this replaces a real compliance determination. If accessibility compliance is a live legal question for your organization, the next step is a conversation with a lawyer or accessibility consultant who can look at your actual video content, audience, and jurisdiction — not another guide.