Accessibility Experts Urge Teams to Test AI

Accessibility practitioners Aaron Gustafson, Jessie Lorenz, and Carie Fisher at Microsoft Build urged teams to embed accessibility early in AI development, warning that AI accelerates existing processes and can scale barriers if accessibility is not integrated. Lorenz noted that fixing accessibility issues later becomes 10x, 100x, and 1000x more expensive as the lifecycle advances.

Accessibility Experts Urge Teams to Test AI At Microsoft Build and in a companion blog post, accessibility practitioners Aaron Gustafson, Jessie Lorenz, and Carie Fisher presented a talk titled "Can Your AI Pass the Accessibility Test?" Aaron Gustafson, June 2026 . They argued that "AI does not fix a broken process; it accelerates whatever process you already have," and warned that AI will scale existing accessibility practices or replicate existing barriers if accessibility is not embedded. Jessie Lorenz outlined a six-stage software pipeline and noted that fixing accessibility issues later becomes far more expensive, citing increases of 10x , 100x , and 1000x as the lifecycle advances. Editorial analysis: This talk frames accessibility as a lifecycle engineering problem that interacts directly with AI tooling and automation. What happened In a Microsoft Build talk and accompanying blog post, accessibility practitioners Aaron Gustafson , Jessie Lorenz , and Carie Fisher presented "Can Your AI Pass the Accessibility Test?" Aaron Gustafson, published 10 Jun 2026 . The presenters laid out a six-stage software pipeline, planning, design, development and coding, code review and CI/CD, public release, and customer feedback, and used it to locate where accessibility work must occur in product development. Jessie Lorenz said the team often observes accessibility issues become 10x , 100x , and 1000x more expensive to fix as they move later in the lifecycle. Technical details The presenters framed the core technical point as process-level: "AI does not fix a broken process; it accelerates whatever process you already have," per the transcript. The talk emphasized that when accessibility is integrated into early stages of design and planning, fixes are a conversation and low-cost; when accessibility is absent from the workflow, AI-driven automation and scale will propagate the same barriers to more users and more surfaces. Editorial analysis: For practitioners, this is a reminder to treat accessibility as an engineering lifecycle concern rather than a post-release checklist. Incorporating accessibility into design systems, CI/CD tests, and code review reduces the risk that generative tools and automation will multiply exclusion. Teams relying on AI for copy, layout, or component generation will need automated checks and accessible design tokens if they want inclusion to scale alongside automation. Context and significance Industry reporting and recent research highlight growing use of generative AI in UX, content generation, and developer tooling. Observed patterns in similar transitions: when automation is introduced into a process without embedding accessibility guardrails, the same usability and assistive-technology failures are reproduced at greater scale. That pattern increases legal, customer-experience, and technical remediation costs across products that serve users with disabilities. What to watch Indicators an observer can track include the adoption of automated accessibility tests in CI/CD pipelines, integration of accessibility constraints into design-token systems, and whether AI-driven authoring tools expose accessibility settings or metadata. Also monitor whether vendor tooling and platform SDKs add first-class accessibility APIs for generated content. Note: The summary above draws on the Microsoft Build talk transcript and Aaron Gustafson's blog post. The authors have not been paraphrased beyond the supplied transcript; where they stated figures or direct lines, those are quoted or attributed to the talk. Scoring Rationale The talk highlights a practical, high-impact risk for AI-enabled product teams: automation can multiply accessibility failures if accessibility is not embedded. This matters to product, design, and engineering teams but is not a frontier-model or regulatory shock. Practice interview problems based on real data 1,500+ SQL & Python problems across 15 industry datasets — the exact type of data you work with. Try 250 free problems /problems