Canada Releases World's First National Standard for Accessible and Equitable AI

On December 3, 2025—International Day of Persons with Disabilities—Canada made history by publishing CAN-ASC-6.2:2025: Accessible and Equitable Artificial Intelligence Systems, the world's first national standard specifically addressing AI accessibility and equity for people with disabilities.

For businesses already navigating ADA compliance, WCAG requirements, and the growing landscape of website accessibility lawsuits, this Canadian standard signals where global accessibility regulation is heading—and why AI accessibility needs to be on your radar now.

Why This Standard Matters for Your Business

With over 4,000 ADA website accessibility lawsuits filed in 2024 alone—and 77% targeting eCommerce websites—accessibility compliance has never been more critical. But here's what many businesses miss: as AI becomes embedded in everything from customer service chatbots to hiring tools to product recommendations, accessibility requirements are expanding beyond websites to include AI systems themselves.

This Canadian standard provides a blueprint for what accessible and equitable AI looks like—and smart businesses will use it to get ahead of inevitable U.S. and international regulations.

What the Standard Requires: The Four Core Principles

The CAN-ASC-6.2:2025 standard establishes that people with disabilities must:

1. Experience equitable benefits from AI systems — AI can't just work for the statistical average user
2. Not experience inequitable harms from AI systems — Including cumulative small harms that compound over time
3. Not experience a loss of rights and freedoms — No discriminatory surveillance or profiling
4. Be given agency and treated with respect — Including the right to choose human alternatives to AI

Key Requirements That Will Impact Your Business

Accessible AI Interfaces

All AI systems, tools, and outputs must meet accessibility standards, specifically CAN-ASC-EN 301 549:2024 (Canada's equivalent to WCAG). This means:

• AI-powered chatbots must work with screen readers
• Voice assistants must offer text alternatives
• AI-generated content must be perceivable, operable, understandable, and robust

Human Alternatives Required

Perhaps the most significant requirement: organizations must provide equivalent human alternatives to AI-assisted decisions. If your AI makes decisions affecting people with disabilities—hiring, customer service, fraud detection, content moderation—you must offer:

• The option for a human decision-maker with relevant expertise
• Human oversight and verification of AI decisions
• No penalties for choosing the human alternative

Real-world example from the standard: If AI sign language interpretation is deployed, Deaf users must be able to request a human interpreter instead, especially in critical settings like healthcare or legal proceedings. Similarly, Blind users facing AI-proctored exams must be able to request human proctors.

Data and Bias Prevention

The standard recognizes that people with disabilities are often statistical outliers—meaning AI systems trained on average users may systematically disadvantage them. Requirements include:

• Preventing bias in training data, data labels, and synthetic data
• Disaggregated performance metrics for people with disabilities
• Continuous monitoring of real-world impacts
• A public registry of harms documenting barriers and inequitable treatment

Transparency and Consent

Organizations must provide accessible, plain-language information about:

• What data was used to train AI systems
• How AI decisions are made
• How to contest or appeal AI decisions
• Who within the organization is accountable for AI decisions

Who Created This Standard?

The standard was developed by a technical committee of leading accessibility and AI experts, including:

• Jutta Treviranus (Chairperson) — Director of the Inclusive Design Research Centre, OCAD University
• Sambhavi Chandrashekar, PhD (Vice-Chairperson) — Global Accessibility Lead, D2L Corporation
• Kate Kalcevich — VP of Innovation, Fable
• Merve Hickok — Founder and Research Director, Center for AI and Digital Policy
• Julia Stoyanovich — Associate Professor and Director, NYU Tandon School of Engineering
• Lisa Snider — Senior Digital Accessibility Consultant, Access Changes Everything Inc.
• Gary Birch — Executive Director, Neil Squire Society
• Clayton Lewis — Professor, University of Colorado Boulder

The committee also included international contributors from the European Disability Forum, the Swedish Disability Rights Federation, Adobe, Statistics Canada, and others.

As Dr. Chandrashekar noted on LinkedIn: "This standard is designed to help organizations and developers design AI systems that are accessible to people with disabilities, ensure fairness and prevent exclusion, establish processes so accessibility and equity remain central throughout the AI lifecycle, and educate others on creating equitable and inclusive AI."

How This Connects to ADA and WCAG Compliance

If you're already working on ADA website compliance or WCAG conformance, this standard extends those principles into AI:

Current RequirementAI Extension Under CAN-ASC-6.2Alt text for imagesAI-generated images must include accessible descriptionsKeyboard navigationAI interfaces must be fully keyboard accessibleScreen reader compatibilityAI outputs must work with assistive technologiesClear languageAI explanations must be in plain languageUser controlUsers must be able to opt out of AI and request human alternatives

What This Means for U.S. Businesses

While this is a Canadian standard, it matters for American businesses for several reasons:

1. If you serve Canadian customers, this standard may apply to you—similar to how New York and California ADA lawsuits target any business serving residents of those states
2. It signals regulatory direction—the EU's European Accessibility Act takes effect June 2025, and U.S. federal agencies are increasingly focused on AI accountability
3. It provides a compliance roadmap—businesses that adopt these practices now will be ahead when similar U.S. requirements emerge
4. It's already referenced in procurement—Canadian government AI procurement will require conformance, affecting any vendor seeking those contracts

Practical Steps to Prepare

For eCommerce and Retail (the most-sued industry for accessibility)

• Audit AI-powered features: search, recommendations, chatbots, checkout assistants
• Ensure AI outputs (product descriptions, sizing recommendations) are accessible
• Provide human customer service alternatives clearly

For HR and Recruiting

• Review AI resume screening and interview tools for disability bias
• Offer human-reviewed application processes as an alternative
• Document accessibility testing of hiring AI

For Healthcare and Financial Services

• Implement human override capabilities for AI decisions affecting benefits, coverage, or access
• Create accessible consent mechanisms for AI data use
• Establish clear accountability chains for AI-assisted decisions

For All Organizations

• Train staff on accessible and equitable AI principles
• Include people with disabilities in AI testing and feedback
• Maintain documentation of AI accessibility conformance

The Bottom Line

With accessibility widget providers facing FTC fines and 25% of 2024's accessibility lawsuits targeting companies using those widgets, the message is clear: quick fixes don't work. Real accessibility requires building it in from the start.

Canada's new AI standard extends this principle to artificial intelligence. As AI becomes embedded in every customer touchpoint, hiring decision, and service interaction, accessibility can't be an afterthought.

The organizations that proactively adopt accessible and equitable AI practices won't just avoid legal risk—they'll serve customers better, access a larger market (61 million Americans live with disabilities), and build AI systems that actually work for everyone.

Resources

• Full Standard: CAN-ASC-6.2:2025 — Accessibility Standards Canada
• CAN-ASC-EN 301 549:2024 — ICT Accessibility Requirements (referenced in the AI standard)
• WCAG 2.1 Guidelines — Web Content Accessibility Guidelines
• Inclusive Design Research Centre — OCAD University (research supporting the standard)
• We Count Initiative — Inclusive AI research

Have questions about how AI accessibility requirements affect your business? Contact TestParty for an accessibility audit that goes beyond surface-level compliance.