Compliance reference · S.O. 2005, c. 11 · O. Reg. 191/11
AODA captions: what Ontario's IASR caption rule means for training video
If your organisation does business in Ontario, has 50 or more employees, and runs internal or external training video, AODA reaches you. The technical artefact the regulation cares about is captions on web content under WCAG 2.0 Level AA — and unlike most US accessibility regimes, AODA pairs that bar with a proactive compliance-reporting cycle that lands in the Ministry's inbox every three years. Operators who run YouTube auto-captioning at the front of the workflow learn this the hard way at report time, when the captions exist but the audit-sampler watching a slice can read the proper-noun failures off the screen.
TL;DR
The Accessibility for Ontarians with Disabilities Act, 2005 (AODA) requires Ontario to be accessible by 2025 — that statutory deadline has now passed. The operative rules live in the Integrated Accessibility Standards Regulation (O. Reg. 191/11), particularly Part II — Information and Communications Standards, sections 11–17. Section 14 binds designated public-sector organisations and large organisations (50+ employees) to make their web content conform to WCAG 2.0 Level AA, with limited exceptions for live captions and pre-2012 archived content. The training-video question is SC 1.2.2: synchronized prerecorded captions at substantial accuracy. Compliance reports are due every three years (next major filing window: 2026 for large private and non-profit organisations under the most recent reporting schedule). The Accessibility Compliance Authority's audit sampling looks the same as ADA Title II's: open the video, watch a slice, judge whether the captions actually convey what the speaker said.
Who AODA reaches
The IASR's Part II uses three tiers of "obligated organisation" — the tier determines which sections apply and on what timeline. For captioning purposes the most consequential are:
- Government of Ontario and the Legislative Assembly. All sections apply, with the earliest implementation timelines (now in steady state).
- Designated public-sector organisations — listed in Schedule 1 of O. Reg. 146/10 — covering universities and colleges, hospitals, public school boards, transit authorities, and most provincial agencies. Section 14 binds them to WCAG 2.0 Level AA on web content since January 2021.
- Large organisations — 50 or more employees in Ontario, private sector or non-profit. Section 14 has bound them to WCAG 2.0 Level AA on new and significantly refreshed web content since January 2021.
- Small organisations — fewer than 50 employees in Ontario. Most of Part II does not apply, but other AODA standards (Customer Service, Employment) still do.
The 50-employee threshold is counted across all of an organisation's Ontario operations, not per-location — a Toronto SaaS company with 30 employees in Toronto, 25 in Ottawa, and an Alberta sales team of 20 has 55 Ontario employees and is a large organisation under the IASR.
Cross-border SaaS companies frequently miss this threshold. A US-headquartered company with a Toronto engineering office of 60 people is a large organisation in Ontario; the internal training catalogue served to the Toronto office is in scope, even if the catalogue is hosted on infrastructure in Virginia.
The technical bar: WCAG 2.0 AA, with practical pressure to 2.1
O. Reg. 191/11 section 14(2) names "the World Wide Web Consortium Web Content Accessibility Guidelines (WCAG) 2.0, at Level AA" as the standard. The regulation has not been updated to 2.1, though the Ministry's plain-language guidance increasingly references 2.1 as good practice and the AODA reform reports (David Onley 2019, Rich Donovan 2023) have recommended a formal update.
For captions specifically, the substantive bar is identical between 2.0 and 2.1: SC 1.2.2 (Captions, Prerecorded) requires captions on every prerecorded audio-visual asset, and SC 1.2.4 (Captions, Live) requires them on live content (with a narrow IASR carve-out). The new 2.1 caption-related criteria sit at AAA and don't bind here. So the captioning bar AODA enforces is the same WCAG bar that ADA Title II enforces — and the same WCAG bar the EAA enforces in Europe.
Operators serving multiple jurisdictions can ship one caption track that meets all three. The compliance and reporting infrastructure is what diverges.
What "training video" means in IASR scope
Section 14 binds "internet websites and web content" — and the IASR's section 14(5) defines "web content" expansively, in line with the W3C definition. Training video falls in scope when it is delivered through a website or web application, which captures essentially every modern LMS-hosted asset and every video served from an organisation's own domain. The narrow carve-out is for "live captions" (SC 1.2.4) and for content posted before January 2012 that has not been "significantly refreshed" since.
Specifically in scope:
- Internal employee training served through an LMS (TalentLMS, Docebo, Cornerstone, Workday Learning, Saba, etc.) where the LMS is web-served.
- Onboarding video on the careers page or candidate portal.
- Customer-academy and product-training video on the public website.
- Lecture-capture video on a university Brightspace, Canvas, Moodle, or Kaltura instance.
- Hospital staff training on the intranet, where the intranet is web-served.
- Compliance-training video on a vendor LMS, where the vendor LMS is what the employer's training is delivered through.
What's outside scope: pre-2012 archived content, live captions on real-time meetings (separate AODA section), and on-device content delivered without the web (rare in practice).
The compliance-reporting rhythm — the bit US operators miss
AODA pairs its WCAG bar with something most US accessibility regimes don't: a proactive compliance reporting cycle. Designated public-sector organisations and large organisations must file an Accessibility Compliance Report (ACR) with the Ministry of Seniors and Accessibility on a recurring schedule — historically every three years for large organisations, with the most recent filing windows in 2014, 2017, 2020, 2023, and the next major filing window in 2026.
The report attests in writing to compliance with each part of the IASR — including section 14's WCAG 2.0 AA bar. False or misleading attestation is a separate offence under section 38 of the Act, with administrative monetary penalties under O. Reg. 432/07.
The operational consequence is that AODA enforcement has a built-in three-year drumbeat. The Accessibility Compliance Authority and Ministry inspectors use the report cycle as their lead source: an organisation that filed "fully compliant" three years ago and now has YouTube auto-captioned video on the website is the obvious investigation target.
Penalties: AMPs scale with organisation size
O. Reg. 432/07 sets out the administrative monetary penalty schedule. For a corporation:
- Major contravention — up to CAD 100,000 per day.
- Moderate contravention — up to CAD 50,000 per day.
- Minor contravention — up to CAD 1,000 per day.
Individual directors and officers face their own scale (up to CAD 50,000 per day for a major contravention). The "per day" structure matters: a finding that the training-video catalogue did not meet WCAG 2.0 AA on the report attestation date can stack penalties for every day the contravention continued.
In practice, the Ministry rarely lands at the top of the AMP scale on first contact — the compliance-and-cooperation track typically gives the organisation a remediation window. But the AMP structure is what backstops the compliance-report drumbeat: filing "compliant" while the catalogue isn't is materially riskier than filing "not yet compliant" with a remediation plan.
How AODA audits actually catch captions
Three patterns catch caption non-compliance:
- Compliance report follow-up. The Ministry samples filings, particularly large-organisation filings that attest full compliance. Inspectors open the public-facing portion of the website, find the training-video catalogue (careers page, customer education, etc.), and watch a slice with captions on. If the captions mangle the content the speaker is teaching, the attestation is in tension with the observed reality.
- Public complaint. The AODA Toolkit routes public complaints into the Ministry's enforcement office. Universities and large public-sector organisations get the bulk of complaints; private-sector complaints typically come from employees, candidates, or customers who hit a specific video and named it in writing.
- Targeted sectoral audit. The Ministry has run sectoral audit pushes (post-secondary, hospitals, retail) where the inspectors land at multiple organisations in a sector with a consistent test methodology. The sampling pattern is the same in each case: open the video, watch a slice, look for whether the captions teach what the speaker is teaching.
The pattern is the same one a 504 OCR investigation or an ADA Title II complaint produces. The compliance bar that gets failed is functional access, and the failure mode is the proper-noun mangling that generic auto-captioning produces with predictable regularity.
The proper-noun failure mode in Ontario training content
Ontario operators who watch the audit-sampling pattern report consistent surfaces:
- Provincial regulatory citations. "PIPEDA" → "pipita"; "PHIPA" → "pip-fa"; "OHSA" → "O-H-S-A" letter-by-letter; "MFIPPA" → splintered fragments. Health-sector and public-sector training is dense with these.
- Ministry programme names. "OHIP" → "owe-hip"; "WSIB" → spelled out; "ServiceOntario" → "service Ontario"; "ARIO" → "Ari-O" (often the agriculture portfolio).
- French terms in bilingual training. Training delivered in Toronto and Ottawa frequently contains French place-names and programme names. "Conseil scolaire de district catholique Centre-Sud" doesn't survive a Whisper-large pass without bias; nor does "L'Université de l'Ontario français".
- Indigenous community names. The training-video terms operators most need to get right are also the ones generic STT has the least training data for: "Anishinaabe", "Haudenosaunee", "Mississaugas of the Credit", "Six Nations of the Grand River". Mangling these in a public-sector cultural-competency module is exactly the failure that draws complaint and inspector attention.
- Healthcare drug and procedure names. The same drug-name mangling pattern that medical training video captions documents applies in Ontario hospital training, plus PHIPA-specific framing the federal HIPAA reference doesn't cover.
The shared pattern: the words an Ontario trainee must learn to do their job are exactly the words generic STT has never seen.
The glossary-biased workflow for AODA-bound organisations
- Build the Ontario glossary once. Provincial regulation acronyms, Ministry programme names, your organisation's internal acronym handbook, the bilingual French terms in your communications glossary, the Indigenous community names from your land-acknowledgement protocol. Connect via Notion, Confluence, or Google Docs, or paste a flat list.
- Process the back-catalogue first. AODA enforcement looks at the catalogue as it stands today, not as it was when it was first captioned. Re-processing the back-catalogue with the glossary baked in clears the most common audit-sampling failures.
- Reviewer pass before publication or LMS upload. The amber-highlight UI shows every glossary-applied term in context. A reviewer can scrub the high-meaning surface in minutes per video, not the hour-plus a hand-fix on YouTube auto-captions takes.
- Document for the Accessibility Compliance Report. Each video gets a row in the asset register: caption file, caption source (vendor + workflow), reviewer, review date. When the next ACR cycle lands, the documentation is the answer to the inspector's question.
AODA compared with ADA Title II, EAA, and Section 504
For an organisation operating across the US-Canada border or into the EU, the regimes layer:
- AODA — Ontario, WCAG 2.0 AA, three-year compliance reporting cycle, AMP-backed, Ministry of Seniors and Accessibility enforcement.
- ADA Title II — US state and local government, WCAG 2.1 AA explicit since 2026-04-24 (large entities), private-suit-driven enforcement.
- Section 508 — US federal procurement, WCAG 2.0 AA, agency contracting officer enforcement.
- Section 504 — US federal-fund recipients, functional access standard, OCR investigations, see section 504 captions reference.
- EAA — EU 27 member states, EN 301 549 V3.2.1 (cross-walks WCAG 2.1 AA), enforceable since 2025-06-28, member-state-specific fine schedules.
Other Canadian provinces are following Ontario's lead — Manitoba's Accessibility Act, British Columbia's Accessible BC Act (in progressive standard development), Nova Scotia's Accessibility Act, and at the federal level the Accessible Canada Act for federally regulated organisations. The cross-walks are not yet fully aligned, but every standard now in force or in development references WCAG.
Related questions
If we're outside Ontario but our customers are in Ontario, does AODA apply?
The IASR binds organisations "providing goods, services or facilities" in Ontario. A US-headquartered SaaS company with no Ontario employees but Ontario customers takes the position that AODA does not bind it directly — but Ontario-based customers themselves are AODA-bound and will require AODA-compliant captions on training video they deliver. In practice, captioning to WCAG 2.0 AA is what your Ontario customers' procurement teams will ask for. The cleaner answer is to caption to that bar globally and avoid the per-jurisdiction fork.
Are auto-captions on YouTube ever sufficient for AODA?
For a low-stakes, no-proper-noun video — a thank-you message with no domain terms — auto-captions can technically meet SC 1.2.2's "captions exist and are reasonably synchronised" bar. For training video with technical content, regulatory citations, or proper nouns, auto-captions virtually always fail the substantive-accuracy reading. The Ministry's compliance-attestation language is "conforms to WCAG 2.0 Level AA," not "has captions of any kind"; an auto-caption track that mangles the speaker's words doesn't actually convey them.
Does AODA require French captions in addition to English?
AODA itself is bilingual but section 14's WCAG bar is per-language: captions must conform to WCAG for whatever languages the original audio uses. A bilingual French-English training programme needs SC 1.2.2-conformant captions in both languages on each video; a unilingual English programme needs them in English. The French Language Services Act may impose additional bilingual requirements on designated public-sector organisations independent of AODA.
What's the next deadline I should plan around?
The most recent IASR amendments (2024 administrative review cycle) have not changed the WCAG 2.0 AA bar or the three-year compliance-report rhythm. The next major large-organisation report filing window is 2026. The structural pressure is the standing AODA Reform recommendations (Donovan 2023) to update the IASR to WCAG 2.1 AA — if and when that lands, the substantive caption bar is unchanged but the Success Criteria around input modalities shift.
Further reading
- WCAG 2.1 AA captions reference
- SC 1.2.2 Captions (Prerecorded) explained
- ADA Title II captions: the 2026-04-24 deadline
- Section 504 captions: federal-fund recipients
- Section 508 captions: federal contractors
- EAA captions — the EU companion rule
- University lecture capture captions
- Why we built GlossCap: the regulatory and operator case