81% of developers trust the accessibility performance of code generated by AI, but data suggests they shouldn't.

Tucson, Arizona, USA, September 30, 2026 / PRNewswire / -- AudioEye, Inc. (Nasdaq ticker: AEYE) (hereinafter referred to as “AudioEye” or “the Company”) is a leading digital accessibility company. Today, the Company released a new study and conducted a business survey, which revealed two key findings: leading AI coding tools failed to create accessible websites, while 81% of teams using these tools believed they were meeting compliance standards.

AI Coding tool has built an accessibility-deficient website with high severity issues

AudioEye has put forward the same requirements to five AI tools – including OpenAI, Anthropic, Google, xAI, and Lovable – to build a news media website, an online store, and a financial services website, and to meet the WCAG 2.2 AA requirements, which are global accessibility compliance standards.

Tests on these 15 websites identified 306 different accessibility issues, which appeared more than 59,000 times on the pages. A single error in the navigation menu or form fields would reappear on every page that used that component. Among the issues discovered, 91% were of moderate to high severity, which means they made it more difficult, or even impossible, for people with disabilities and AI agents to perform tasks. These types of issues are also those that consistently appear in digital accessibility lawsuits.

These websites written by LLM have an average of 55 questions per page, while typical websites in the AudioEye "2026 Digital Accessibility Index" have an average of 62 questions per page. The requirement for AI tools to build accessible websites has not made much of a difference; the accessibility level of these pages is still almost as poor as that of average websites today.

AI doesn't "hallucinate" about these failures. It was trained in a network environment that is already lacking in accessibility features, so it can write code with accessibility defects just as quickly as the team releases content.

Trust Gap: Teams are full of confidence in the code generated by AI, but complaints are on the rise

AudioEye also investigated the level of trust developers and business leaders have in the code written by AI. The results showed a gap between their perceptions and their actual experiences. Although 81% of respondents said they were confident that the code generated by AI on the website met accessibility standards, 73% reported an increase in accessibility issues and complaints since the adoption of AI.

Nearly half (46%) of the respondents received legal letters or lawsuits, and among them, 71% stated that the pages that received complaints used the AI tool.

What does this mean for teams that use AI to build products?

This does not mean that AI is incompatible with accessibility, but rather that tools trained to identify accessibility issues in web content are unable to judge or fix such problems on their own. Organizations that rely solely on LLM to handle accessibility will accumulate legal risks with each newly published page.

AudioEye CEO Kelly Georgevich stated: "The speed at which AI generates code is faster than ever before. Clearly, there is a misplaced sense of security among people that AI tools can produce barrier-free and compliant code. AudioEye Intelligence is our proprietary dataset, built on over a million manual reviews, billions of interactions, and years of auxiliary technology testing. We combine this intelligence with the latest generative AI to enable our customers to release barrier-free code on a large scale."

Methodology

AudioEye requires five AI tools (OpenAI, Anthropic, Google, xAI, and Lovable) to each build three websites that meet the WCAG 2.2 AA standards. Subsequently, AudioEye conducted an accessibility audit and testing on these 15 websites, checking issues such as keyboard focus, dialog box handling, form error messages, and screen reader output. Each issue was then rated as high, medium, or low severity based on its impact on end-users and legal risks.

About AudioEye

AudioEye is committed to ensuring that the digital future we build is accessible to people with disabilities and AI agents. Our comprehensive solutions, serving over 129,000 clients, combine industry-leading AI automation technology with expert repairs supported by information provided by the disability community. This powerful combination offers industry-leading litigation protection and AI agent readiness capabilities. Our solutions are protected by 25 U.S. patents, including 24/7 accessibility monitoring, automatic WCAG issue testing and repair, expert testing, developer tools, and legal protection, helping organizations to confidently create accessible digital experiences for everyone.

Media Contact

Aulani Capuchin

Investor Contact

Tom Colton