CAN-ASC-6.2:2025 Compliance Assessment

Executive Summary

Overall Compliance: 70-85% (Moderate to Strong)

PAICE demonstrates moderate to strong compliance with CAN-ASC-6.2:2025, the Canadian standard for AI accessibility. Strong performance in interface accessibility and transparency, with opportunities for improvement in AI-specific features like explainability and alternative interaction modes.

CAN-ASC-6.2:2025 is Canada's first national standard specifically addressing accessibility requirements for artificial intelligence systems. It extends beyond traditional web accessibility to address AI-specific concerns including algorithmic transparency, explainability, bias mitigation, and alternative interaction modes.

Standard Context

Published: 2025 by Standards Council of Canada
Scope: AI systems used in Canada, particularly those affecting accessibility
Relationship: Complements WCAG 2.1 with AI-specific requirements
Status: Voluntary standard (may become regulatory requirement)

Compliance by Category

Category 1: Interface Accessibility (95%)

Status: Strong Compliance

WCAG 2.1 AA compliance at 99%
Keyboard navigation fully supported
Screen reader compatibility verified
Touch targets meet 44x44px minimum
Color contrast exceeds 4.5:1 ratio

Category 2: Transparency & Documentation (90%)

Status: Strong Compliance

AI usage clearly disclosed in assessment
Methodology documented in whitepaper
Limitations clearly stated
Privacy practices transparent
Contact information readily available

Category 3: Explainability (70%)

Status: Moderate Compliance

✓ Score breakdown provided by dimension
✓ Confidence intervals explained
✓ Methodology accessible in whitepaper
⚠ Limited real-time explanation of AI decisions
⚠ No interactive "why this score" feature yet

Planned: Q2 2026 - Enhanced explainability features

Category 4: Alternative Interaction Modes (65%)

Status: Moderate Compliance

✓ Text-based interface (primary mode)
✓ Keyboard-only operation supported
✓ Screen reader compatible
⚠ No voice input option yet
⚠ No simplified/plain language mode yet

Planned: Q3 2026 - Voice input and simplified mode

Category 5: Bias Mitigation (85%)

Status: Strong Compliance

Assessment measures collaboration skills, not personal characteristics
No demographic data collected
Scoring based on observable behaviors only
Regular bias testing planned
Diverse testing group for validation

Category 6: User Control (80%)

Status: Strong Compliance

Users control when to start assessment
Can pause and resume (session management)
Can request data deletion
Can export results in multiple formats
Clear consent process before assessment

Category 7: Error Prevention & Recovery (75%)

Status: Moderate Compliance

✓ Clear error messages with guidance
✓ Form validation with helpful feedback
✓ Session recovery after interruption
⚠ Limited ability to correct AI misinterpretations
⚠ No "undo" for submitted responses yet

Planned: Q2 2026 - Response editing feature

Detailed Compliance Matrix

Requirement	Status	Notes
6.2.1 Interface meets WCAG 2.1 AA	✓ Pass (99%)	Exceeds requirement
6.2.2 AI usage disclosed	✓ Pass	Clear disclosure in assessment
6.2.3 Methodology documented	✓ Pass	Whitepaper available
6.2.4 Limitations stated	✓ Pass	Research preview status clear
6.2.5 Decisions explainable	⚠ Partial	Score breakdown provided; enhanced features planned
6.2.6 Alternative modes available	⚠ Partial	Text/keyboard; voice planned Q3 2026
6.2.7 Bias testing conducted	✓ Pass	Ongoing testing program
6.2.8 User control provided	✓ Pass	Start, pause, export, delete
6.2.9 Error recovery supported	⚠ Partial	Session recovery; editing planned Q2 2026
6.2.10 Privacy protected	✓ Pass	GDPR/CCPA compliant
6.2.11 Feedback mechanism	✓ Pass	Multiple channels available
6.2.12 Continuous improvement	✓ Pass	Regular updates and testing

Gaps and Remediation Plan

Gap 1: Enhanced Explainability

Current State: Score breakdown by dimension provided, but limited real-time explanation of specific AI decisions.

Target State: Interactive "why this score" feature with detailed reasoning for each dimension.

Timeline: Q2 2026 - Enhanced explainability dashboard

Gap 2: Voice Input Mode

Current State: Text and keyboard input only.

Target State: Voice input option for users who prefer or require speech-based interaction.

Timeline: Q3 2026 - Voice input implementation

Gap 3: Simplified Language Mode

Current State: Standard language level (Grade 10-12 reading level).

Target State: Optional simplified/plain language mode (Grade 6-8 reading level).

Timeline: Q3 2026 - Plain language mode

Gap 4: Response Editing

Current State: Responses cannot be edited after submission.

Target State: Users can review and edit responses before final submission.

Timeline: Q2 2026 - Response review and editing feature

Strengths and Differentiators

PAICE's AI Accessibility Strengths

Built-in Accessibility: Accessibility considered from initial design, not retrofitted
Transparent AI Usage: Clear disclosure of AI role in assessment process
Bias-Aware Design: Assessment measures skills, not personal characteristics
User Control: Users maintain control over data and assessment process
Continuous Improvement: Regular testing and updates based on user feedback
Public Documentation: Methodology and limitations openly shared