AI Accessibility Guidelines
Creating inclusive learning and work environments where AI enhances accessibility and supports Universal Design for Learning
Accessibility means ensuring all learners, including students with disabilities and other challenges to learning, can access and engage with all course materials, activities, and assessments.
When thoughtfully employed, generative AI can function as a powerful accessibility resource that enhances learning opportunities for all students through Universal Design for Learning (UDL) principles.
Our Accessibility-First Approach: Use AI to expand options and create inclusive experiences, never to restrict access or create barriers.
Creating Accessible Content with AI
Text Accessibility Prompts
"Rewrite this assignment description to be more accessible for students with reading difficulties"
"Break this complex reading assignment into smaller, manageable chunks with clear learning objectives"
"Generate a plain-language summary of this technical concept that maintains academic accuracy"
"Create a glossary of technical terms from this document with simple definitions"
Visual Content Prompts
"Create detailed alt-text descriptions for these course images that explain both visual elements and educational purpose"
"Describe this chart/graph in a way that conveys the same information to someone who cannot see it"
"Convert this visual process diagram into step-by-step text instructions"
"Generate audio descriptions for this video that explain visual elements"
Quality Control Reminder
Always review AI-generated accessibility features for accuracy and alignment with WCAG standards. AI is a starting point, not the final answer.
AI Tool Accessibility Evaluation
Basic Accessibility Checks
- Keyboard-only navigation capability
- Screen reader compatibility
- High contrast and magnification support
- Clear, consistent interface design
- Captioning for video-based tools
Advanced Accessibility Features
- Voice control and dictation support
- Customizable fonts and spacing
- Multiple input methods (touch, voice, keyboard)
- Focus indicators and logical tab order
- Alternative text for all interface elements
Evaluation Resources
Faculty Considerations
Mindset Shifts for AI Accessibility
- Think "accessible by design" rather than retrofitting accessibility later
- Use AI to expand options, never to restrict or eliminate accommodations
- Consider AI as one tool in the UDL toolkit, not a complete solution
- Test AI tools with assistive technology before recommending them
Collaboration & Support
- Connect with the Office for Disability Equity early and often
- Include accessibility in course AI policies, not as an afterthought
- Ask for feedback from students who use AI accessibility features
- Share successful AI accessibility strategies with colleagues
Avoiding Accessibility Barriers
When AI Concerns Arise, Don't:
- • Force assessments back to in-person only
- • Eliminate flexible deadline options
- • Remove alternative format submissions
- • Create blanket AI restrictions without exceptions
Consider Students Who Need:
- • Assistive technology format conversions
- • Distraction-free testing environments
- • Extra time to process information
- • AI tools as accommodation support
Based on guidance from Cornell University's Center for Teaching Innovation
Privacy & Safety Guidelines
Never Share
- •Specific disability diagnoses or medical conditions
- •Accommodation letters or documentation
- •Medical records or therapy notes
- •Detailed descriptions of personal limitations
Safe Alternatives
- •Use general requests: "Please simplify this text"
- •Ask for accessibility features without disclosure
- •Request multiple format options as good practice
- •Focus on the task rather than the reason
FERPA and Disability Information
Under FERPA, disability-related educational records receive the same protection as other student records. Use UM's Amplify GenAI for privacy-protected AI assistance.
Read Full Privacy GuidelinesCurrent Research & Evidence
UCF Project RAISE
The University of Central Florida, with United Cerebral Palsy of Central Florida, developed 'ZB' (Zoobee)—an AI-driven socially assistive robot helping elementary students with disabilities develop social skills and coding abilities.
University of Illinois Speech Accessibility Project
Partnering with Microsoft, Google, Amazon, Meta, and Apple to improve voice recognition for people with diverse speech patterns and disabilities, achieving 18-60% accuracy improvements.
AI Writing Support Research
A 2025 study with 124 students with disabilities found AI particularly beneficial for students with ADHD, dyslexia, dyspraxia, and autism, primarily using chatbots and rewriting applications.
Vanderbilt Planning Assistant
Vanderbilt University's AI-powered Planning Assistant scans course syllabi to extract key dates and add them to student calendars, with future plans for assignment task breakdown.
Key Finding: Students expressed strong desire to participate in AI policymaking and for universities to provide generative AI training.
Quick Reference: Getting Started
Faculty Essentials
- Test AI tools with assistive technology first
- Use AI to create accessible content proactively
- Ensure course AI policies don't create barriers
- Connect with Office for Disability Equity (ODE)
- Review AI-generated features for accuracy
Student Essentials
- Start with UM's Amplify GenAI platform
- Focus on learning enhancement, not work completion
- Test AI tools with your assistive technology
- Work with Office for Disability Equity (ODE) on accommodations
- Use general accessibility requests for privacy
Staff Essentials
- Evaluate AI tools for accessibility compliance
- Include accessibility in selection criteria
- Provide accessible AI training materials
- Monitor and report accessibility barriers
- Collaborate with Office for Disability Equity (ODE)
Standards & Compliance
Accessibility Support & Resources
UM is committed to ensuring all community members can effectively use AI tools.