What is this course about?
Conversational AI interfaces break accessibility assumptions that static pages do not: output arrives token by token, focus moves unpredictably, and screen readers announce partial content. This course covers conformance for streaming interfaces, cognitive accessibility as an access requirement, and producing the conformance evidence procurement will demand.
Who is this course for?
- Accessibility coordinators and 508 program staff
- Front-end developers building agency interfaces
- Digital service teams
- Contracting staff evaluating conformance claims
What do I need before starting?
- Familiarity with WCAG or Section 508 requirements
- Some front-end development context helpful
What will I be able to do afterwards?
- Apply Section 508 obligations to conversational AI interfaces
- Make streaming output behave correctly with screen readers
- Manage keyboard navigation and focus in a chat interface
- Treat cognitive accessibility as an access requirement
- Evaluate an Accessibility Conformance Report critically
What does each module cover?
How does Section 508 apply to a chat interface?
40 minMapping the obligations onto an interface type the standards did not anticipate.
Objectives
- Apply 508 obligations to conversational interfaces
- Identify where the standards are ambiguous
- Set the conformance target
Topics
Activity. Map 508 requirements onto an existing agency chat interface.
What happens when a screen reader meets streaming text?
60 minThe core technical problem — output arriving token by token while a screen reader tries to announce it.
Objectives
- Explain how live regions behave with streaming output
- Configure announcements that are useful rather than chaotic
- Test with multiple screen readers
Topics
Activity. Test streaming output with two screen readers and fix the announcement behavior.
How do you manage keyboard navigation and focus?
50 minFocus management in an interface where content appears continuously.
Objectives
- Design a keyboard navigation model for chat
- Manage focus as new content arrives
- Provide skip and navigation affordances
Topics
Activity. Navigate your interface entirely by keyboard and fix every trap and dead end.
Why is cognitive accessibility an access requirement?
45 minReading level, consistency, and predictability as conformance concerns rather than nice-to-haves.
Objectives
- Treat cognitive load as an accessibility requirement
- Control reading level and consistency in generated output
- Design predictable interaction patterns
Topics
Activity. Measure and reduce cognitive load in your interface's output.
How do you handle errors and waiting accessibly?
45 minLoading states, errors, and timeouts announced properly rather than silently.
Objectives
- Announce loading and processing states
- Convey errors accessibly and actionably
- Handle timeouts without stranding a user
Topics
Activity. Test error and timeout states with a screen reader and remediate.
How do you read an Accessibility Conformance Report?
45 minEvaluating vendor conformance claims critically rather than accepting the document.
Objectives
- Read an ACR critically
- Identify claims that testing would disprove
- Require evidence rather than assertion
Topics
Activity. Evaluate a real ACR and test three of its claims.
How do you test with real users?
50 minTesting with assistive technology users, which finds what automated tools never do.
Objectives
- Recruit and compensate assistive technology testers
- Structure sessions that produce actionable findings
- Prioritize findings by impact
Topics
Activity. Run a session with an assistive technology user and prioritize the findings.
Conducting the conformance audit
50 minThe workshop module: a complete audit with evidence and a remediation plan.
Objectives
- Conduct a full conformance audit
- Produce evidence for each finding
- Build a prioritized remediation plan
Topics
Activity. Complete the audit and produce the remediation plan.
What is the capstone project?
508 conformance audit of an AI interface
Conduct a complete accessibility audit of an agency AI interface covering streaming announcements, keyboard navigation, cognitive load, error states, and real assistive technology user testing — with evidence and a prioritized remediation plan.
Deliverable: An audit report with evidence, user testing findings, and a remediation plan.
How are learners assessed?
- Audit must include testing with a real assistive technology user
- Keyboard navigation verified with no traps or dead ends
- Findings prioritized by user impact rather than ease of fix
What ships with the course?
Facilitator guide
Session-by-session running order, discussion prompts, and the questions that reliably derail a room.
Learner workbook
Exercises, checklists, and the templates each module's activity produces.
Hands-on lab environment
A sandboxed ibl.ai deployment so exercises run against real agents, not screenshots.
Assessment bank
Scenario questions and rubric criteria mapped to each stated learning outcome.
Source bibliography
Every primary regulation and standard cited on this page, linked and dated.
Which AI agents does this course use?
The hands-on modules run against agents already deployable on the ibl.ai platform for government.
Where does the course material come from?
Every module is grounded in primary sources — the regulation, standard, or research itself, not a summary of it. Each was resolved at authoring time.
- Section508.gov
GSA
Federal accessibility obligations and conformance reporting requirements.
- Web Content Accessibility Guidelines
W3C
The technical conformance criteria applied throughout.
- Federal Acquisition Regulation
Acquisition.gov
Accessibility requirements in federal acquisition.
- AI Guide for Government
GSA Centers of Excellence
Federal practice guidance for accessible AI services.
Delivery notes
Binding guidance for anyone preparing and delivering this course.
- Module 2 is the technically hardest and most valuable content in the course. Streaming announcement behavior differs between screen readers and versions; test across at least two and document the differences.
- Assistive technology users must be recruited and paid for Module 7. Unpaid participation is both unfair and produces worse findings, and agencies have mechanisms to compensate testers.
- Automated accessibility tools will pass interfaces that are unusable. Show this explicitly — run a tool against a deliberately broken interface and let it pass.
- Module 6's ACR evaluation should use a real report. Vendors routinely claim conformance that testing disproves, and participants need to see it once.
- Coordinate with GOV-3 — accessibility is a component of citizen service design and the two courses should share the audit methodology.
Why run AI training on a platform you own?
You own the course, not a licence to it
Course content, learner data, and the platform run inside your perimeter — you own all the code and the data.
Model-agnostic delivery
Run the course's AI components on any LLM — Claude, GPT, Llama, Gemini, Command — and switch anytime.
No per-seat training licences
Usage-based or self-hosted, so cost tracks actual use rather than headcount.
Deploy anywhere
Cloud, private VPC, on-premise, or fully air-gapped — including for cohorts that cannot use public AI tools.
Frequently asked questions
What does the Section 508 and Accessible AI Interfaces course cover?
Conversational AI interfaces break accessibility assumptions that static pages do not: output arrives token by token, focus moves unpredictably, and screen readers announce partial content. This course covers conformance for streaming interfaces, cognitive accessibility as an access requirement, and producing the conformance evidence procurement will demand. It runs 5 hours across 8 modules across 8 modules, at intermediate level, and closes with a capstone: 508 conformance audit of an AI interface.
Who should take Section 508 and Accessible AI Interfaces?
It is written for Accessibility coordinators and 508 program staff, Front-end developers building agency interfaces, Digital service teams, Contracting staff evaluating conformance claims. Prerequisites: Familiarity with WCAG or Section 508 requirements; Some front-end development context helpful.
Can we run this course on our own infrastructure?
Yes. ibl.ai is model-agnostic and deploy-anywhere — cloud, private VPC, on-premise, or fully air-gapped — and you own all the code and the data. Cohort data, submissions, and any material learners upload stay inside your perimeter, which matters for government teams that cannot send work to a public AI tool.
How do we get access to Section 508 and Accessible AI Interfaces?
Request access and we will set it up for your cohort — hosted by ibl.ai, or running against your own deployment. Tell us the group size and timing you need, and whether it should run inside your own perimeter.
How much does AI training for government cost on ibl.ai?
There is no per-seat pricing — you pay for usage or self-host and pay only for the infrastructure, so a 5,000-person rollout does not cost 5,000 licences. 1.6M+ users across 400+ organizations run the platform this way, including NVIDIA, MIT, and Syracuse University.