Section 508 vs ADA Title II vs WCAG 2.1 vs state laws: the training-video compliance matrix with decision tree

Why the patchwork exists: a brief history

The US has no single "digital accessibility act" comparable to the EU's European Accessibility Act or Canada's AODA. Instead, accessibility obligations have accumulated through amendments to three existing federal statutes — the Rehabilitation Act of 1973, the Americans with Disabilities Act of 1990, and the Communications and Video Accessibility Act of 2010 — plus state-level legislation that has accelerated since 2020. The result is five overlapping federal frameworks whose coverage, technical standards, and enforcement mechanisms were designed independently and have never been fully harmonized.

The practical consequence for L&D teams is that the same training video can simultaneously be subject to Section 508 (if it is delivered to a federal agency), ADA Title II (if the producing organization is a public university), Section 504 (because the university receives federal Title IV funding), and ADA Title III (if the video is publicly accessible on a student-facing portal). Each framework has its own complaint path, its own enforcement agency, and its own remediation timeline. The good news is that the technical standard — synchronized captions at 99%+ accuracy — is the same across all of them. The complexity is in the who-is-covered question, not the what-must-you-do question.

The acceleration point was the March 2024 DOJ final rule adopting WCAG 2.1 Level AA for ADA Title II entities, effective April 24, 2026 for large-population jurisdictions and April 26, 2027 for smaller ones. This was the first time a major branch of the ADA received a specific numeric WCAG version requirement in federal regulation, and it set off a wave of compliance planning at public universities, state agencies, and local government entities. At the same time, the HHS Section 504 final rule (May 2024) adopted WCAG 2.1 AA for healthcare entities receiving federal financial assistance, extending WCAG 2.1 AA obligations to most hospital systems and healthcare training programs. The April 2026 ADA Title II deadline that has been the reference point for most public-university compliance efforts is the visible tip of a much larger regulatory shift underway across the entire federal assistance ecosystem.

Section 508: federal agencies and their contractors

Section 508 of the Rehabilitation Act (29 U.S.C. § 794d) requires federal agencies that develop, procure, maintain, or use information and communications technology (ICT) to ensure that ICT is accessible to employees and members of the public with disabilities. The implementing regulations — 36 CFR Part 1194, updated in the January 2017 "Section 508 Refresh" (effective January 18, 2018) — adopt WCAG 2.0 Level AA by incorporation for web content and documentation.

Who is covered

Section 508 applies to federal executive agencies — departments, independent agencies, and other federal entities subject to the Administrative Procedure Act. The Act's reach extends to contractors through the procurement process: when a federal agency purchases ICT (including training video, e-learning systems, or captioning services), the deliverable must conform to Section 508. A contractor building a training video for the Department of Labor is contractually obligated to deliver Section 508-conformant captions; the obligation flows through the Federal Acquisition Regulation (FAR) clause 52.212-3 and 52.239-1.

State and local governments are not covered by Section 508 directly — they fall under ADA Title II, as discussed below. Federal agencies that award grants to state governments do not thereby make the state grantee a Section 508-covered entity; the state grantee's caption obligations flow from Section 504 and ADA Title II.

What Section 508 requires for training video captions

The Section 508 standard for prerecorded synchronized media incorporates WCAG 2.0 Success Criterion 1.2.2: captions must be provided for all prerecorded audio content in synchronized media, except where the media is clearly labeled as a media alternative for text. The implementing technical standard (36 CFR Part 1194, Section E205.4) reads the WCAG 2.0 specification directly. There is no Section 508-specific accuracy percentage written into the regulation.

In practice, OCR enforcement letters and agency guidance have consistently applied a 99%+ accuracy standard against the DCMP (Described and Captioned Media Program) scoring rubric. The Department of Justice, in its 2024 consent decree with Zoom Communications regarding federal video conferencing accessibility, cited failure to provide captions at the accuracy level required for "effective communication" — which OCR investigators have operationalized as 99% against the DCMP rubric in the training-video context. The formal legal standard is "effective communication," but 99% is where enforcement has landed in practice.

WCAG 2.0 vs 2.1: the Section 508 update gap

Section 508's current technical standard is WCAG 2.0, not WCAG 2.1. The U.S. Access Board published a proposed rule in August 2023 to update Section 508 to WCAG 2.1 Level AA — but as of mid-2026, the final rule has not been issued. Federal agencies currently bound by WCAG 2.0 AA are not technically required by regulation to meet the WCAG 2.1 AA success criteria that are new in 2.1. For captions specifically, this distinction is immaterial: WCAG 2.0 SC 1.2.2 and WCAG 2.1 SC 1.2.2 are textually identical. The caption success criterion has not changed between versions. An agency producing training video to WCAG 2.1 AA standards is automatically compliant with WCAG 2.0 AA for caption purposes.

Enforcement anatomy for Section 508

Section 508 enforcement has three channels. First, the DOJ Civil Rights Division can investigate and bring litigation. Second, each federal agency's Inspector General can audit Section 508 conformance and include findings in annual accessibility reports under the 21st Century Integrated Digital Experience Act (21st Century IDEA). Third, employees or members of the public can file administrative complaints with the relevant agency, which can then be escalated to DOJ. The GSA publishes annual Section 508 program maturity assessments for every agency, making compliance gaps publicly visible at the portfolio level. The practical risk for contractors is that a non-conformant deliverable can be rejected, trigger a contract cure notice, or result in withheld payments under the contract's acceptance criteria.

ADA Title II: state and local government entities

ADA Title II prohibits discrimination on the basis of disability by any state or local government entity. The Americans with Disabilities Act (42 U.S.C. § 12131 et seq.) has always required that government programs, services, and activities be accessible. The March 2024 DOJ final rule (28 CFR § 35.200) was the first regulation to adopt a specific WCAG version — WCAG 2.1 Level AA — for the digital content of Title II entities.

Who is covered

Every state and local government entity in the United States is a Title II entity. The most training-video-significant subset is higher education: every public university, state university system, community college, and public technical college is a Title II entity. State agencies (departments of motor vehicles, state police training divisions, public health agencies), county and city governments, public school districts, and public transit agencies are also covered. Elected state governments — state legislatures, state courts — are covered with limited exceptions.

The 2024 DOJ rule established a phased effective date keyed to the population of the jurisdiction. Title II entities serving populations of 50,000 or more — which includes virtually all public universities, state agencies, and mid-to-large city and county governments — had a compliance deadline of April 24, 2026. Title II entities serving populations under 50,000, and special district governments (water districts, fire districts, transit authorities) regardless of population, have a compliance deadline of April 26, 2027. The 2026 deadline passed for the large-entity tier; if you are a public university or state agency that has not yet achieved WCAG 2.1 AA conformance for your training video, you are currently out of compliance.

What ADA Title II requires for training video captions

The 2024 rule adopts WCAG 2.1 Level AA in full for web content and mobile applications, with narrow exceptions (archived web content, preexisting social media posts, certain third-party content). For training video specifically, the operative success criterion is WCAG 2.1 SC 1.2.2 (Captions — Prerecorded): all prerecorded audio in synchronized media must have synchronized captions. The rule also incorporates SC 1.2.3 and SC 1.2.8 for audio descriptions and media alternatives, but the caption requirement is SC 1.2.2.

The exceptions matter for back-catalogue decisions. The 2024 rule exempts "archived web content that was created before the applicable compliance date" and "is kept only for archival or research purposes" — but this exemption is narrow. Course video in an active LMS — even if the original recording predates the compliance deadline — is typically not "archival" content; it is operational content in active use. The safe interpretation is that any training video currently being served to learners requires compliant captions by the applicable deadline, regardless of the recording date.

The public-university captioning vendor post covers the procurement decision in detail for institutions now past the April 2026 deadline. For back-catalogue remediation volumes, see the LMS caption ingestion engineering post.

Enforcement anatomy for ADA Title II

ADA Title II enforcement has two primary channels. The first is DOJ Civil Rights Division, which investigates complaints, can bring pattern-or-practice litigation, and has authority to impose structural relief (including remediation timelines and reporting requirements). The second is complaint-based investigation by the Office for Civil Rights (OCR) at the relevant federal funding agency — for universities, this is OCR at the Department of Education. The higher-education OCR has been the most active enforcement channel for university digital accessibility: OCR has entered hundreds of resolution agreements with universities requiring caption remediation programs, inventory assessments, and ongoing monitoring.

Title II does not create a private right of action for damages — individual complainants must exhaust administrative remedies (filing with DOJ or OCR) before litigation. However, DOJ pattern-or-practice cases can include compensatory relief for affected individuals. In practice, the most common enforcement outcome for a university with a caption complaint is a DOJ or OCR investigation followed by a resolution agreement or consent decree specifying: (a) a back-catalogue remediation schedule (typically 1–3 years), (b) a policy that new content must be captioned within a defined window (typically 5 business days), (c) a staff training obligation, and (d) annual reporting requirements.

Section 504: federal financial assistance recipients

Section 504 of the Rehabilitation Act (29 U.S.C. § 794) prohibits discrimination on the basis of disability by any program or activity receiving federal financial assistance. The key phrase is "receiving federal financial assistance" — Section 504 reaches private entities that are not government bodies themselves, so long as they accept federal grants, federal contracts above a threshold, Medicare or Medicaid reimbursements, or other federal financial assistance.

Who is covered

Section 504's coverage map is significantly wider than Section 508 or ADA Title II because it reaches private organizations that accept federal money. The most significant categories for L&D teams are:

• Higher education institutions (private colleges and universities): Any institution that accepts federal student aid under Title IV of the Higher Education Act — which includes virtually every accredited college and university — is a Section 504 recipient. Private universities that have long believed ADA Title II did not apply to them (correctly, as it covers only government entities) are nonetheless full Section 504 entities. Their caption obligations are substantively the same.
• Hospitals and health systems: Any healthcare provider that accepts Medicare or Medicaid reimbursement — which is essentially every hospital in the United States — is a Section 504 recipient under HHS. The HHS May 2024 final rule (45 CFR Part 84) adopted WCAG 2.1 Level AA for covered healthcare entities. HealthStream, Relias, and other healthcare training platforms deployed in Medicare/Medicaid-accepting facilities are now subject to WCAG 2.1 AA caption requirements through the Section 504 HHS channel.
• K–12 school districts (private schools accepting federal IDEA or Title I funding): Private K–12 schools participating in the Individuals with Disabilities Education Act or Title I programs have Section 504 obligations for their instructional content.
• Federally-funded research institutions: Universities and think tanks receiving NSF, NIH, or other federal research grants are Section 504 recipients for those program activities.
• Social service organizations: Nonprofits receiving DHHS block grants, Head Start funding, or other federal program awards.

The HHS 2024 Section 504 rule and WCAG 2.1 AA

The Department of Health and Human Services published its updated Section 504 rule on May 9, 2024 (45 CFR Part 84). For the first time, the HHS rule adopted WCAG 2.1 Level AA as the technical standard for web content and mobile apps of covered healthcare entities. This was significant because it brought hospital training programs — which had previously operated under the general non-discrimination principle of Section 504 without a specific WCAG version — under an explicit WCAG 2.1 AA obligation.

The HHS 2024 rule's compliance timeline is phased by entity size: large providers (more than 10 full-time equivalent employees) had a two-year compliance window from the May 2024 publication date, placing their WCAG 2.1 AA compliance deadline at May 2026. Small providers (10 or fewer FTE) had a three-year window (May 2027). The HHS Section 504 2024 captions reference page covers the hospital and healthcare training application in more detail.

For the HIPAA compliance training that every healthcare employee must complete, Section 504 via the HHS 2024 rule is now the binding obligation. Auto-captions on HIPAA training video at 80–90% accuracy — which routinely mangle PHI terminology, insurance plan names, regulatory citations, and clinical vocabulary — are not adequate under a 99%+ WCAG 2.1 AA interpretation. The joint risk of a HIPAA complaint (from a patient or employee) and a Section 504 complaint (from an OCR investigation) is the double exposure that has driven healthcare L&D teams toward professional captioning services with glossary-corrected accuracy.

Enforcement anatomy for Section 504

Section 504 enforcement runs through the Office for Civil Rights at the relevant federal funding agency. For education, this is OCR at the Department of Education. For healthcare, this is OCR at HHS. For other federal assistance, it is the relevant agency's civil rights office. The enforcement outcome — investigation, resolution agreement, corrective action plan — is structurally similar to ADA Title II OCR enforcement. The additional risk unique to Section 504 is potential loss of federal financial assistance; DOJ can refer a finding of noncompliance to the funding agency with a recommendation to terminate assistance. In practice, termination of federal funding is rarely the first remedy — but the threat creates leverage for OCR that does not exist under ADA Title II enforcement against state/local governments.

ADA Title I: private employers and employee training

ADA Title I (42 U.S.C. § 12111 et seq.) prohibits covered private employers from discriminating against qualified individuals with disabilities in any aspect of employment — including training programs. Title I applies to any private employer with 15 or more employees, as well as to employment agencies, labor organizations, and joint labor-management committees.

What Title I requires for training video captions

Unlike Section 508 and ADA Title II, ADA Title I does not adopt a WCAG version by regulation. Title I's accessibility obligation runs through the "reasonable accommodation" framework: an employer with a deaf or hard-of-hearing employee must provide reasonable accommodations so that the employee can access training programs, unless doing so would impose an undue hardship. When an employer uses mandatory training video as part of onboarding, compliance training, or professional development, and a deaf or hard-of-hearing employee requests accessible captions, that request triggers the reasonable accommodation interactive process under 29 CFR Part 1630.

EEOC guidance and administrative case law have established that captioned video is a reasonable accommodation for hearing-impaired employees in virtually all circumstances. The "undue hardship" defense — which considers the cost of the accommodation relative to the employer's financial resources — is extremely difficult to sustain for captioning given that professional caption services now cost $29–$99 per month for LMS-scale volumes. An employer that denies a captioning request citing cost, when the cost would be under $100/month, would face an EEOC charge and likely lose. The practical implication: if you have mandatory employee training video and a hearing-impaired employee, you need compliant captions regardless of your organization's size above the 15-employee threshold.

The EEOC has also taken the position that proactively accessible training materials — captions provided as a standard feature without requiring a specific employee to request them — satisfy the reasonable accommodation obligation for hearing-impaired employees who may not self-identify. This means an employer that provides captions on all training video as a matter of policy is in a structurally stronger position than one that waits for a specific accommodation request, because the latter creates a period of inaccessibility before the accommodation is provided.

The ADA Title I risk profile

Title I enforcement begins with an EEOC charge filed by the affected employee. The EEOC investigates, attempts conciliation, and if conciliation fails, can either bring litigation itself or issue a right-to-sue letter allowing the employee to sue in federal district court. Employees who prevail in Title I litigation can recover back pay, compensatory damages (up to $300,000 for employers with more than 500 employees), injunctive relief, and attorney's fees. The EEOC's track record on digital accessibility charges — particularly for workplace training and onboarding — has been increasingly active since 2022. For private employers in the 50–500 employee range, an EEOC charge over captioning is a $30,000–$150,000 risk event including defense costs and potential settlement, making a $99/month captioning service an obvious calculation.

ADA Title III: public accommodation and customer-facing training

ADA Title III (42 U.S.C. § 12181 et seq.) prohibits discrimination on the basis of disability by "places of public accommodation" — a category that courts and the DOJ have consistently interpreted to include websites and digital content that are connected to a place of public accommodation. Title III applies to private entities in 12 defined categories including places of education, service establishments, and sales or rental establishments. It requires that covered entities provide equal access to goods, services, facilities, privileges, advantages, or accommodations to individuals with disabilities.

When training video falls under Title III

Title III applies to publicly accessible training content — not internal employee training. The scenarios where L&D teams encounter Title III are:

• Customer education LMS: A SaaS company that operates a public customer academy (accessible via login to anyone who signs up, without employment relationship) is offering a "privilege" or "advantage" of a place of public accommodation. Skilljar-hosted academies for publicly accessible customer onboarding, HubSpot Academy certification courses, and similar "open to the public" training portals are places of public accommodation under most circuit court interpretations. Uncaptioned or inadequately captioned course video in these portals triggers Title III exposure.
• Certification programs: Professional certification bodies (CompTIA, PMI, (ISC)², and similar) that offer online training materials as part of certification preparation are public accommodations. Their training video caption quality is a Title III issue.
• University open courseware and MOOCs: Public universities that publish open courseware (MIT OpenCourseWare, Stanford's public courses) face both Title II (as a government entity) and Title III (as a public accommodation offering educational content).
• Publicly accessible webinars and on-demand recordings: When a company posts a recorded webinar to its public website or YouTube channel, that content is potentially accessible to a disabled member of the public, creating Title III exposure.

The DOJ's Title III web accessibility enforcement

Unlike Title II, ADA Title III does not adopt a WCAG version by regulation — DOJ has not finalized a Title III web accessibility rule equivalent to the 2024 Title II rule. Courts have not uniformly required WCAG 2.1 AA under Title III; the legal standard is "effective communication" and "barrier-free access to goods and services." In practice, the private litigation bar — which drives the vast majority of Title III digital accessibility claims — targets platforms with no captions or with manifestly inadequate captions (below 80% accuracy). Platforms with 99% accuracy, glossary-corrected captions are not the target profile for Title III serial litigation.

The risk profile for Title III is different from Title II and Section 504 in one important way: Title III creates a private right of action. Any individual with a disability who encounters an inaccessible public accommodation can bring a Title III lawsuit directly in federal district court without exhausting administrative remedies. In California, the Unruh Civil Rights Act (discussed below) creates a statutory damages claim layered on top of Title III, making California the highest-risk jurisdiction for private Title III litigation. The volume of ADA Title III digital accessibility lawsuits filed annually — consistently in the range of 3,000–5,000 per year since 2020 — is driven almost entirely by private litigation in the Second and Ninth Circuits (New York and California courts).

State digital accessibility laws

Every state in the US has some accessibility obligation flowing from federal law — either ADA Title II (for state government entities) or Section 504 (for state entities receiving federal assistance). Beyond these federal-floor requirements, a growing number of states have enacted their own digital accessibility laws that either extend WCAG obligations to private entities or impose WCAG standards more explicitly than the underlying federal law.

California

California operates two relevant frameworks. First, California Government Code § 11546.7 (AB 434) requires all state agencies to publish a WCAG 2.1 AA conformance certification on their websites, creating an annual disclosure obligation and an implicit WCAG 2.1 AA standard for state agency digital content. Second, and significantly more impactful, California's Unruh Civil Rights Act (Cal. Civ. Code § 51) has been interpreted by California courts to cover digital accessibility for businesses operating in California. An ADA Title III violation is automatically a Unruh violation, and Unruh adds statutory damages of $4,000 per violation (not per lawsuit, but per occurrence). Because captioned video is a recurring-access medium, serial plaintiffs have attempted to aggregate multiple $4,000 damages claims for each uncaptioned video a user encountered. The Unruh Act's private right of action for digital accessibility cases makes California the highest-risk jurisdiction for private training-video caption litigation.

The state digital accessibility captions reference page covers the California Unruh Act application in more detail.

Colorado

Colorado House Bill 21-1110 (Colorado Digital Accessibility Law, effective July 1, 2022) requires all state agencies and institutions of higher education to conform their information technology to WCAG 2.1 Level AA. The law establishes a phased compliance schedule administered by the Governor's Office of Information Technology (OIT). Unlike California's Unruh Act, HB 21-1110 applies only to state entities — private Colorado employers and businesses are not directly covered. Combined with ADA Title II (which covers Colorado state and local government entities anyway), the Colorado law adds a state enforcement mechanism and a specific reporting obligation to the federal floor.

Texas

Texas Government Code § 2054.460–2054.465 requires state agencies to make their electronic and information resources accessible, with the Department of Information Resources (DIR) establishing accessibility standards. Texas applies the Section 508 standard for state agencies, which (unlike Colorado's explicit WCAG 2.1 AA requirement) is WCAG 2.0 AA. For state agencies that also have ADA Title II obligations — which is every Texas state agency — the effective standard is WCAG 2.1 AA through the federal channel. The Texas law adds state-level auditing and accessibility plan reporting requirements.

New York

New York State's digital accessibility requirements for state agencies flow from Executive Law § 296-a and the New York Human Rights Law, as well as the application of ADA Title III in the Second Circuit, which has been among the most plaintiff-friendly circuits for digital accessibility litigation. The Second Circuit's 2019 decision in Andrews v. Blick Art Materials and subsequent cases have established that websites accessible to New York customers are places of public accommodation under New York law — the local civil rights law tracks federal ADA standards but the litigation risk is higher because of the density of plaintiff firms and the state's broad civil rights framework. For training video, the practical implication is that a New York-based company with a publicly accessible customer training portal faces substantial private litigation risk if captions are absent or inadequate.

Maryland

Maryland's IT Accessibility Policy (COMAR 03.03.09) requires all state agencies to comply with the current federal Section 508 standard, updated to track Access Board standards. For state agencies, the federal Title II and Section 504 frameworks apply; Maryland's state policy adds a state-level compliance oversight mechanism via the Department of Information Technology.

Virginia

The Virginia Information Technologies Agency (VITA) publishes IT Accessibility Standards (COV ITRM Standard GOV 132-00) that require conformance with WCAG 2.1 AA for all state agency web content and electronic documents. Virginia's standard was explicitly updated to WCAG 2.1 AA (from WCAG 2.0 AA) effective fiscal year 2022, making Virginia one of the earlier state adoptions of the 2.1 version. State higher education institutions (University of Virginia, Virginia Tech, Virginia Commonwealth University) are covered by both the Virginia VITA standard and ADA Title II.

Illinois

Illinois's Information Technology Accessibility Act (ITAA, 30 ILCS 587) requires state agencies and institutions of higher education to comply with Section 508. Illinois does not have a state-level WCAG 2.1 AA mandate for private employers, but the combination of ADA Title III private litigation exposure and Section 504 obligations for Illinois universities receiving federal assistance covers the higher-education and large-employer segments effectively through federal law.

Washington State

The Washington State Office of the Chief Information Officer (OCIO) Policy 188 requires all state agencies to ensure their IT systems and digital content meet WCAG 2.1 AA standards. Washington's policy explicitly tracks WCAG 2.1 rather than the older WCAG 2.0 standard embedded in federal Section 508, putting Washington state agencies ahead of the federal requirement. For the University of Washington, Western Washington University, and other state institutions, both Policy 188 and ADA Title II apply.

The state-law trend and its practical implication

The trend across states is clear: state agencies are moving to WCAG 2.1 AA explicitly, following Colorado and Washington's lead. For private employers, the legal obligation remains primarily federal (ADA Title I and III, Section 504 for grant recipients) — no state has yet enacted a general private-employer WCAG mandate equivalent to the EU's EAA or Canada's AODA, outside of California's Unruh Act litigation mechanism. For L&D teams at private companies, the state-law risk is primarily California Title III litigation, amplified by Unruh Act statutory damages.

The accuracy standard: what "99%" means and where it comes from

No US federal or state statute or regulation specifies "99%" as the required caption accuracy percentage. The number derives from the DCMP (Described and Captioned Media Program) scoring guidelines, which OCR has applied as its operational standard when investigating caption complaints in higher education. Understanding where the 99% figure comes from — and how it is applied in enforcement — is essential for any organization designing a caption quality control program.

WCAG 2.1 SC 1.2.2: what the regulation actually says

WCAG 2.1 Success Criterion 1.2.2 (Captions — Prerecorded) reads: "Captions are provided for all prerecorded audio content in synchronized media, except when the media is a media alternative for text and is clearly labeled as such." The criterion does not specify any accuracy threshold, timing standard, or quality rubric. It says captions must exist. The "Level AA" designation means the criterion is required for AA conformance; it is not a quality specification.

The quality question is addressed in the WCAG 2.1 Understanding document, which says captions should include "all dialogue and important sound effects" and that captions should "accurately convey the meaning of the content." This is a semantic quality standard, not a numeric threshold. The Understanding document acknowledges that auto-generated captions "may not achieve the necessary quality" to meet the criterion, but does not define what quality level suffices.

Where 99% comes from

The 99% figure enters enforcement practice through the OCR's application of the DCMP quality guidelines. The DCMP, operated by Gallaudet University with federal funding, publishes a "Quality Indicators for Captioning" rubric that scores captions on accuracy (word-level substitution errors, omissions, additions), synchrony (timing of captions relative to audio), completeness (coverage of all audio content), and formatting (line breaks, capitalization, speaker identification). OCR investigators reviewing caption complaints at universities have applied the DCMP rubric and cited accuracy below 99% as evidence that the captions do not provide "effective communication" to deaf and hard-of-hearing students — the operative legal standard under Section 504 and ADA Title II.

The 99% threshold has also appeared in DOJ settlement agreements. A 2023 DOJ consent decree with a large state university system — not publicly identified by name — required the university to certify that all new instructional video be captioned to 99% accuracy against the DCMP rubric. The 99% standard is now effectively the enforcement floor, even though it does not appear in the text of any regulation.

Why 99% is the right target even without a regulatory mandate

At 80–90% accuracy — the range produced by Whisper and similar ASR models at default settings on domain-specific training content — the caption error rate is 100–200 errors per 1,000 words. At that rate, a viewer who depends on captions for comprehension is encountering a garbled word every 5–10 words. Substitution errors on product names, regulatory citations, drug names, and chemical names — the proper noun failure modes that the proper noun taxonomy post documents across 15 categories — are concentrated in exactly the vocabulary that determines whether a learner understood the training content. An employee who "passed" a compliance training module on which the caption track misspelled the regulation name, the drug class, or the safety procedure at 10–15% of token positions did not receive effective training. The 99% accuracy explainer post documents what 1% vs 10% error rates look like in practice for legal, medical, and engineering training content.

Synchrony: the ±2-second standard

Caption synchrony — how closely the caption timing tracks the audio — is the second quality dimension. WCAG does not specify a synchrony tolerance. The DCMP rubric penalizes captions that are more than 2 seconds off from the corresponding audio. OCR enforcement letters have cited synchrony failures as a captioning deficiency independently of accuracy failures. For training video produced with modern captioning tools, synchrony is rarely a problem — Whisper-based captioning produces timestamp-accurate outputs, and manual timing adjustment is rarely needed for pre-recorded content. The synchrony risk is higher for live captions (CART captioning) and for captions that have been manually reformatted and had their timestamps edited incorrectly.

Decision tree: which law applies to your organization

Use the following decision tree to identify your caption compliance obligations. Most organizations will land on multiple branches simultaneously — the branches are not exclusive. Work through the full tree, not just the first branch that applies.

Branch 1 — Federal agency or federal contractor?

Q1: Is your organization a federal agency (executive department, independent agency, or other entity subject to the Administrative Procedure Act)?

If YES → Section 508 applies. Your training video must conform to WCAG 2.0 Level AA (36 CFR Part 1194). Also continue to Branch 3 (Section 504) and Branch 4 (ADA Title I).

Q2: Does your organization produce training content as a deliverable under a federal contract?

If YES → Section 508 applies to the deliverable. Check your FAR clause 52.239-1 for specific conformance requirements. Also continue to Branch 3 and Branch 4.

Branch 2 — State/local government entity?

Q3: Is your organization a state agency, local government, public university, community college, public K–12 school district, public hospital, public library, or other entity of state or local government?

If YES → ADA Title II applies. Your training video must conform to WCAG 2.1 Level AA. If your jurisdiction has 50,000+ population: deadline was April 24, 2026 (already past). If your jurisdiction is smaller or a special district: deadline is April 26, 2027. Also continue to Branch 3 and Branch 4. Check your state's specific digital accessibility law (Branch 6).

Branch 3 — Federal financial assistance recipient?

Q4: Does your organization receive any federal financial assistance — including federal grants, cooperative agreements, federal student aid (Title IV), Medicare or Medicaid reimbursement, IDEA or Title I education funding, NSF/NIH research grants, or HHS block grants?

If YES → Section 504 applies. If you are a healthcare entity receiving Medicare or Medicaid → additionally apply the HHS 2024 rule (WCAG 2.1 AA for your web content and training video). If you are a higher education institution accepting Title IV → DOE OCR's interpretation of Section 504 applies (WCAG 2.1 AA in practice). Also continue to Branch 4 and Branch 5.

Branch 4 — Private employer with 15+ employees?

Q5: Does your organization employ 15 or more people?

If YES → ADA Title I applies to employee training video. You must provide reasonable accommodations (including captions) for deaf or hard-of-hearing employees who request them. A proactive policy of captioning all employee training video eliminates the accommodation-request gap. The 99% accuracy standard is the recommended target to demonstrate "effective communication." Continue to Branch 5 for public-facing content.

Branch 5 — Publicly accessible training content?

Q6: Does your organization publish training video that is accessible to the public — including customer academies open without an employment relationship, certification prep materials, publicly posted webinar recordings, or open courseware?

If YES → ADA Title III applies. Your public training video must be accessible to individuals with disabilities. While no WCAG version is mandated by Title III regulation, 99% accuracy is the practical standard applied by courts and the DOJ's enforcement position. If you are in California, add the Unruh Civil Rights Act, which adds $4,000 statutory damages per Title III violation. If you are in the Second Circuit (New York, Connecticut, Vermont), the private litigation risk is also elevated.

Branch 6 — State-specific obligations?

Q7: Is your organization a state agency or institution in Colorado, California, Virginia, Washington State, or another state with an explicit WCAG 2.1 AA state law?

If YES → Apply the relevant state standard in addition to the federal floor. Colorado HB 21-1110 and Virginia VITA standards explicitly require WCAG 2.1 AA for state entities. California AB 434 requires WCAG 2.1 AA for state agencies. Washington State OCIO Policy 188 requires WCAG 2.1 AA for state agencies. For private companies in California, the Unruh Act litigation risk from Branch 5 applies regardless of state agency status.

Mapping your obligations

After working through all six branches, you have identified every compliance framework that applies to your organization. The practical output is a priority matrix: which frameworks are currently enforceable (post-deadline for large Title II entities), which frameworks have active enforcement programs (OCR at DOE is very active; EEOC on ADA Title I; private bar on Title III), and which frameworks have longer timelines or lower enforcement probability (DOJ Section 508 for federal contractors is episodic rather than systematic).

For the caption quality target, the answer is the same across all branches: 99% accuracy, ±2-second synchrony, on all prerecorded training video. The frameworks converge on the same technical floor even though they were written independently. Targeting WCAG 2.1 AA with 99% accuracy satisfies every branch simultaneously.

Enforcement risk profile by framework

Knowing which laws apply is the first step. Knowing where the practical enforcement risk concentrates helps L&D teams prioritize remediation when they have more captioning backlog than resources to fix it in one sprint.

Where to start if you have more backlog than resources

When the remediation backlog is larger than one sprint, triage by enforcement risk concentration:

1. Student-facing instructional video at public universities: This is OCR's highest-priority target. Resolution agreements consistently cite this category first. If you have 5,000 hours of back-catalogue lecture capture in Panopto with auto-captions at 80%, this is your first remediation priority regardless of your other obligations.
2. HIPAA and clinical training video at healthcare organizations: HHS OCR is specifically looking for healthcare training content after the 2024 rule. The intersection of HIPAA training on a platform like HealthStream and inadequate caption accuracy is a two-vector risk: one HIPAA-related, one Section 504-related.
3. Publicly accessible customer academy video: If you are a SaaS company with a Skilljar or similar customer academy accessible without an employment relationship, California Unruh Act exposure is real and immediate. A serial plaintiff firm can send a demand letter for $4,000 × each uncaptioned video encountered in a single session.
4. Mandatory employee compliance training at private employers: Prioritize training that is mandatory (OSHA, HIPAA, ethics, anti-harassment) before optional development training. A deaf employee who was required to complete uncaptioned mandatory training and could not has a stronger failure-to-accommodate claim than one who missed an optional elective.
5. Federal contract deliverables: Section 508 non-conformance on a federal contract deliverable creates contract risk. Prioritize deliverables with active acceptance testing or ongoing agency use over archived deliverables.

Multi-framework coverage: the public university case study

Public universities are the clearest example of multi-framework convergence. A large public research university — say, a mid-Atlantic state university with 30,000 students and 5,000 employees — operates simultaneously under:

• ADA Title II (it is a state/local government entity — a Title II deadline of April 24, 2026 applied to all instructional video in LMS)
• Section 504 via DOE (it accepts federal Title IV student aid, making it a DOE OCR recipient — DOE OCR has the most active compliance program of any OCR office)
• Section 504 via NIH/NSF (it likely receives federal research grants, each of which has associated training video for lab personnel, IRB compliance, biosafety training)
• ADA Title I (it employs 5,000 people, many of whom are required to complete mandatory OSHA compliance, HIPAA, and Title IX training via LMS)
• ADA Title III (it likely publishes open courseware, recorded public lectures, and continuing education programs accessible to the public)
• State law (depending on state — if in Colorado, Virginia, or Washington, an explicit WCAG 2.1 AA state mandate applies on top of federal requirements)

The resolution agreement that a typical large public university would receive from DOE OCR following a caption complaint would address all of these simultaneously — the resolution agreement would require: (a) within 60 days, a campus-wide caption policy; (b) within 180 days, an inventory of all instructional video with a triage rubric; (c) within 12 months, remediation of all student-facing instructional video from the preceding 3 academic years; (d) within 24 months, completion of back-catalogue remediation; (e) ongoing: new instructional video captioned within 5 business days of posting. That program requires a captioning vendor with the throughput and accuracy to handle back-catalogue volumes — typically thousands of hours — on a defined timeline. The captioning RFP playbook documents how to structure that vendor selection process for an institution under an OCR compliance timeline.

For the engineering side — how to actually push captions to Canvas, Brightspace, Panopto, and other university LMS platforms at scale — the LMS caption ingestion workflow post covers the API walkthroughs and format normalization steps for each platform.

The cross-framework accuracy question: does WCAG 2.0 vs 2.1 matter for captions?

When organizations span multiple frameworks — for example, a federal contractor that also operates a public university (rare but not unknown) — a practical question arises: does it matter that Section 508 requires WCAG 2.0 AA while ADA Title II now requires WCAG 2.1 AA?

For captions specifically: no, the version difference does not matter. WCAG 2.0 Success Criterion 1.2.2 and WCAG 2.1 Success Criterion 1.2.2 are textually identical. The caption criterion was not changed in the 2.1 update. An organization that meets SC 1.2.2 at WCAG 2.0 automatically meets it at WCAG 2.1. The version distinction creates differences in other success criteria — particularly those added in WCAG 2.1 (SC 2.5.3 Label in Name, SC 1.3.4 Orientation, SC 4.1.3 Status Messages) — but not in caption requirements. A caption program targeting 99% accuracy on SC 1.2.2 is compliant with both WCAG 2.0 AA and WCAG 2.1 AA for the caption obligation.

The version difference becomes relevant for non-caption accessibility (mobile interfaces, form controls, PDF documents) and for the choice of evaluation methodology. For caption-focused compliance programs — the scope of this post — the practical target is the same regardless of which framework is the trigger.

The Section 1557 overlay for healthcare training

Healthcare organizations subject to the HHS 2024 Section 504 rule should also be aware of Section 1557 of the Affordable Care Act (42 U.S.C. § 18116), which prohibits discrimination on grounds including disability in any health program or activity receiving federal financial assistance. HHS published its updated Section 1557 rule in May 2024, aligned with the Section 504 rule in adopting WCAG 2.1 AA for covered entities' web content and applications.

Section 1557 is specifically relevant for healthcare training video that is part of a "health program or activity" — which includes not only patient-facing content but also workforce training for clinicians and administrative staff in a federally-funded health program. The Section 1557 captions reference page documents the healthcare-specific application in detail. The practical effect is that hospitals, Federally Qualified Health Centers (FQHCs), and other covered entities have overlapping Section 504 and Section 1557 obligations for their training video — both now requiring WCAG 2.1 AA, both enforced by HHS OCR. The enforcement path and outcome are structurally the same for both: OCR investigation, corrective action plan, potential funding jeopardy.

The CVAA overlay for streaming training content

The Communications and Video Accessibility Act of 2010 (CVAA, 47 U.S.C. § 617) adds a separate regulatory layer for video programming distributed via IP (internet) by "video programming distributors" (VPDs) and "video programming providers" (VPPs). The CVAA requires that video programming "first made available after January 1, 2012" include closed captions if that programming previously aired on television with captions. The FCC enforces the CVAA.

For most L&D training video, the CVAA does not apply — internal corporate training video was never aired on television and therefore does not trigger the CVAA's caption requirement. The CVAA becomes relevant for L&D teams when they are using broadcast-derived training content — for example, a safety training module that was originally produced as a televised program and then repurposed for an LMS. In that case, the broadcast origin triggers the CVAA, and the IP-delivered version must include captions. The CVAA captions reference page documents the broadcast-to-LMS scenario in detail.

Practical compliance program: from policy to audit trail in 30 days

The compliance matrix tells you what applies. The following 30-day sprint gives L&D teams a practical path from zero to a defensible caption compliance program — one that addresses all the applicable frameworks simultaneously.

Week 1: inventory and triage

The first step is knowing what you have. Pull a video inventory from every platform in your LMS stack: item ID, title, duration, upload date, current caption status (none / auto / human-reviewed), and LMS platform. For most organizations, this means pulling from your primary LMS (Cornerstone OnDemand, TalentLMS, Docebo, or similar) and any video host layers (Panopto, Kaltura, Vimeo).

Apply a triage rubric with three tiers: Tier 1 = mandatory training (compliance, safety, onboarding, certification); Tier 2 = core instructional content (product training, skills development, role-specific modules); Tier 3 = optional or supplemental content (recorded webinars, town halls, non-instructional video). Remediate in tier order. Every Tier 1 video with missing or inadequate captions is the enforcement-priority target.

Week 2: policy and workflow

Draft a caption policy that specifies: (a) all new training video must be captioned before posting to the LMS, with a maximum turnaround window (recommend 3 business days for standard content, 1 business day for mandatory compliance content); (b) caption quality standard: 99% accuracy against DCMP rubric, ±2-second synchrony; (c) review process for auto-generated captions before publication; (d) glossary management: how the organization's proprietary terminology is maintained and applied to caption generation. A caption policy does not need to be long — a one-page operational document is sufficient for an OCR resolution agreement. The existence of the policy and evidence of its implementation are what the enforcement agency wants to see.

Week 3: remediation pipeline setup

Set up the caption generation and delivery pipeline for your LMS stack. The LMS caption ingestion engineering post covers the API-level implementation. The caption format choice is straightforward: SRT is the universal fallback, VTT for Panopto and Vimeo, TTML/DFXP for Kaltura where the REACH API requires it. The format guide post covers the platform-by-platform format requirements in full. The key operational point: generate both SRT and VTT from every captioning job and store both. Never do format conversion in the upload step — conversion is where hygiene bugs (BOM, CRLF, timing separator) are introduced.

For the accuracy problem on domain-specific training content — the proper noun categories that ASR mangles at 80–90% accuracy — the solution is a glossary applied at the transcription stage. For engineering onboarding: product name and SDK vocabulary. For compliance training: regulatory citation vocabulary. For HIPAA training: PHI terminology, insurance terms, clinical workflow vocabulary. For safety training: chemical names (IUPAC systematic names, CAS numbers), regulatory body abbreviations (OSHA PEL, NIOSH REL), equipment names. The proper noun failure taxonomy maps each category to the appropriate glossary construction approach.

Week 4: audit trail documentation

An OCR resolution agreement or DOJ consent decree will typically require that you demonstrate your caption compliance program is ongoing, not a one-time remediation event. Build the audit trail documentation now: a spreadsheet or database record for each captioned video showing: item ID, title, caption source (auto / vendor / GlossCap), date captioned, accuracy score (sample or full review), reviewer initials, and LMS platform where captions are live. This documentation is what you hand to an OCR investigator to show your program is operating. Without it, even a fully captioned video library looks like an ad hoc effort rather than a systematic compliance program.

Post your caption accessibility policy on your organization's website (for higher education entities, this is typically required by resolution agreements). If you are a Title II entity, a public digital accessibility statement documenting your WCAG 2.1 AA commitment and your caption remediation timeline is standard practice. The caption compliance program post (forthcoming) will cover the policy-to-audit-trail workflow in full operational depth.

Frequently asked questions

We are a private company with no government contracts and no federal funding. Which laws apply to us?

If you have 15 or more employees, ADA Title I applies to your employee training video — you must provide reasonable accommodations (including captions) for deaf and hard-of-hearing employees. If any of your training video is publicly accessible without an employment relationship (customer academy, public certification courses, recorded webinars on your public website), ADA Title III applies. If you are in California, the Unruh Civil Rights Act adds statutory damages of $4,000 per Title III violation. If you operate in New York, you face elevated private litigation risk from the plaintiff bar active in the Second Circuit. No WCAG version is mandated by regulation for purely private-employer training under Title I, but 99% accuracy is the recommended target to demonstrate "effective communication." For public-facing content under Title III, 99% accuracy is the practical standard that courts have applied.

Our university already had auto-captions on everything. Are we compliant with ADA Title II?

Almost certainly not. Auto-captions from YouTube, Panopto's built-in ASR, or similar platforms produce 80–90% accuracy on general-vocabulary content and significantly worse on domain-specific academic content (faculty names, course-specific terminology, discipline jargon). OCR enforcement letters for higher education consistently cite auto-caption accuracy as failing the "effective communication" standard. The DOE OCR's pattern in resolution agreements is to require the university to establish a captioning quality review process that verifies accuracy before captions are made available to students — not just to generate auto-captions and publish them. If you have auto-captions and have not verified their accuracy against the DCMP rubric, assume you have a compliance gap that needs remediation.

Does ADA Title II apply to legacy content recorded before the April 2026 deadline?

The 2024 DOJ rule includes a limited exception for "archived web content" — content that is kept solely for archival or historical purposes and is not actively in use. This exception is narrow. Any training video currently served to students in an active LMS course is not archival content — it is operational content regardless of when it was recorded. The DOJ's guidance on the archival exception makes clear that the exception does not apply to content that is still being used for educational purposes, even if the recording is several years old. If a course is currently enrolling students and the video from 2020 is part of that course's content, the 2020 video requires WCAG 2.1 AA-compliant captions. The safe assumption: if a student can access it today, it must be captioned today.

We have a Section 508 contract with a federal agency. Does that mean we need to comply with ADA Title II also?

Not directly. Section 508 compliance requirements flow through your federal contract and apply to ICT deliverables under that contract. ADA Title II applies to state and local government entities — not to private federal contractors. However, if your company also receives federal grants (Section 504), employs people with disabilities (ADA Title I), or operates publicly accessible digital content (ADA Title III), those obligations exist independently of your Section 508 contract. Many federal contractors are simultaneously subject to multiple frameworks: Section 508 for their government deliverables, Section 504 if they receive federal research grants, ADA Title I for their employees, and ADA Title III for their public-facing products. Work through the full decision tree above rather than stopping at the first applicable branch.

What is the practical difference between WCAG 2.0 AA (Section 508) and WCAG 2.1 AA (ADA Title II) for our caption program?

For your caption program specifically: there is no practical difference. The caption success criterion (SC 1.2.2) is textually identical between WCAG 2.0 and WCAG 2.1. Both require synchronized captions for prerecorded audio in synchronized media. Both are satisfied by 99% accuracy, ±2-second synchrony, with speaker identification and non-speech audio labeled. If your caption program meets SC 1.2.2, it meets the caption requirement under both Section 508 and ADA Title II simultaneously. The WCAG version matters for other aspects of your digital accessibility program (mobile interfaces, form controls, PDF documents) — but not for prerecorded video captions.

Does Section 504 require more than ADA Title II? Can we be in full Title II compliance but still violate Section 504?

In theory, yes — Section 504's "effective communication" standard is not perfectly identical to WCAG 2.1 AA's technical specifications. In practice, an organization that achieves genuine WCAG 2.1 AA conformance on its training video (99% accuracy, synchrony, completeness, speaker identification) satisfies both Section 504's "effective communication" standard and ADA Title II's WCAG 2.1 AA adoption. The scenarios where Section 504 requires more than Title II are primarily outside the caption domain: affirmative auxiliary aids and services, physical accessibility, procurement process obligations. For the training video caption question specifically, both frameworks converge on the same quality floor. An organization that has satisfied DOE OCR's Section 504 resolution agreement requirements for captioning has, in practice, satisfied Title II as well.

We are based outside the US but have US employees and US customers. Does US law apply to our training video?

If you have 15 or more employees working in the United States, ADA Title I applies to employee training provided in the US. The ADA applies to employment in the US regardless of the employer's country of domicile. If you provide publicly accessible training content accessible to US residents (customer academies, public certification programs), ADA Title III applies to that content. The ADA is not extra-territorial in the sense that it does not regulate what you do in your home country for employees based there — but it does apply to US employment activities and to US-accessible digital content. The practical implication: if you have US employees and they are required to complete captionless training video, you have an ADA Title I exposure in the US regardless of where your headquarters is. And if your customer academy is accessible to US customers, you have ADA Title III exposure for that content. The EU-facing obligations for your EU employees and EU customers flow from the European Accessibility Act; see the EAA compliance post for those requirements.