Error-Prone AI Accessibility Tools Reportedly Lead to Navigation Issues for Blind Internet Users

AI-powered accessibility overlays on websites frequently mislabel or misinterpret content, in turn complicating navigation for blind users and others with disabilities. Users report that the AI tools interfere with screen readers and mislead them with inaccurate descriptions. The reported unreliability in these tools have prompted legal action, as the companies behind them seek compliance with accessibility laws.

Risk classification

• Primary risk domain: 1 Discrimination & Toxicity
• Primary risk subdomain: 1.3 Unequal performance across groups

The AI overlays perform poorly specifically for disabled users who rely on assistive technologies, resulting in unequal digital access and a degraded user experience compared to sighted users.

Additional risk subdomains

• 7.3 Lack of capability or robustness: The AI systems exhibit a clear lack of capability by generating inaccurate image descriptions and failing to function reliably under standard web conditions.

Causal factors

• Entity: AI
• Intent: Unintentional
• Timing: Post-deployment

The failures and subsequent lawsuits were caused by the deployed AI overlays performing unreliately and generating incorrect descriptions after being integrated into websites.

EU AI Act risk tier

• Risk tier: 2 High Risk

High Risk: The AI systems have significant implications for fundamental rights, specifically the right to equal digital access for disabled individuals, and are deployed to comply with statutory accessibility regulations.

AI system and alleged parties

• AI system: AI-driven software, EqualWeb overlay, UserWay (EqualWeb, UserWay)
• AI purpose: Accessibility Assurance; Image Description
• Behaviour type: Tool
• Alleged developer: UserWay, EqualWeb, Developers of AI-based accessibility tools
• Alleged deployer: Zara, Pemex, LVMH, Companies using AI-based accessibility tools, Capita
• Alleged harmed parties: Visually impaired people, Jakob Rosin, Blind people

Harm severity

Severity ratings are based on this scale.

Highest direct severity in any category: Minor. Severity is scored from Negligible to Catastrophic in each harm category, for harm the reports describe as caused directly or indirectly by the AI system.

• Physical: direct Negligible, indirect Negligible
• Infrastructure: direct Negligible, indirect Negligible
• Property: direct Negligible, indirect Negligible
• Financial: direct Negligible, indirect Substantial
• Environmental: direct Negligible, indirect Negligible
• Malicious content: direct Negligible, indirect Negligible
• Differential treatment: direct Minor, indirect Minor
• Civil rights: direct Minor, indirect Minor
• Democracy: direct Negligible, indirect Negligible
• Privacy: direct Negligible, indirect Negligible
• Psychological: direct Negligible, indirect Negligible
• Epistemic: direct Negligible, indirect Negligible
• Child sexual exploitation and abuse: direct Negligible, indirect Negligible

Financial

Reported: Yes, the report notes that over 900 companies faced legal action and lawsuits over accessibility failures, though specific monetary losses are not quantified.

Indirectly caused: Significant legal defense fees, settlement costs, and compliance penalties for the 900+ companies facing lawsuits due to unreliable AI overlays.

Inferred additional harm: Millions of dollars in cumulative legal expenses, fines, and the subsequent costs of rebuilding websites with manual accessibility solutions.

Differential treatment

Reported: Yes, the report explicitly describes a failure to provide equal digital access for disabled people.

Directly caused: The AI overlays directly caused a degraded, slower, and confusing browsing experience for blind users by misinterpreting web elements and interfering with screen readers.

Indirectly caused: Disabled users were effectively excluded from fully participating in online activities, such as viewing event schedules or shopping, due to unreliable automated tools.

Inferred additional harm: Widespread unequal treatment and digital exclusion of millions of disabled individuals across the 360,000+ websites utilizing these flawed AI overlays.

Civil rights

Reported: Yes, the report describes claims of alleged failure to provide equal digital access in violation of the Americans with Disabilities Act (ADA).

Directly caused: The failure of the AI systems directly resulted in non-compliance with civil rights regulations, violating disabled users' rights to equal access.

Indirectly caused: Companies violated civil rights laws by relying on automated overlays instead of implementing genuine, manual accessibility accommodations.

Inferred additional harm: Systemic civil rights violations affecting thousands of disabled individuals who were denied equal access to public and private digital services.

People affected

• Occurrences reported: 1
• People reportedly harmed: 900
• People reportedly exposed: 900

Potential causes

Management

• Prioritizing Speed Over Safety: Companies chose cheap, fast overlays over proper bespoke accessibility.
• Misleading Marketing Practices: Software companies promised absolute legal protection to boost sales.

Technology

• Unreliable Computer Vision: AI misidentified bullet point image as a toilet due to lack of accuracy.
• Assistive Tech Interference: Overlays slowed down screen readers and disrupted normal navigation.
• Immature AI Capability: AI technology is not yet ready to handle complex web structures.

Data Inputs

• Flawed Web Structural Design: Poor website structure made it difficult for AI tools to parse content.
• Lack of Image Context: AI lacked context to differentiate bullet points from actual objects.

Human Factors

• Overreliance on Automation: Web operators assumed automated software required no manual intervention.
• Misplaced Trust in Vendor Claims: Buyers believed vendor promises of 100 percent compliance.

Process and Methods

• No Manual Verification: Websites deployed AI overlays without any human-in-the-loop review.
• Inadequate Accessibility Testing: Failure to test the overlay with actual screen reader users.

Regulatory Environment

• Strict Accessibility Laws: Impending regulations like EU 2025 laws created pressure for fast fixes.
• Rise in Accessibility Lawsuits: Legal threats under the ADA drove panic buying of quick-fix software.

Information quality

• Classification confidence: High
• Reason for confidence: The report provides clear, factual accounts of the AI overlays' failures, specific user experiences, and statistical data on the resulting lawsuits, allowing for highly confident classifications.
• Ambiguities identified: The exact technical architecture of the overlays is not detailed, but their functional failures are clearly described.

AI-powered accessibility overlays deployed on commercial websites failed to provide reliable navigation for disabled users, leading to increased legal liabilities and civil rights concerns. The incident represents a commercial software and compliance failure with negligible national security implications, though it highlights challenges in automated civil rights compliance.

• Overall national security impact: Minor
• Scope: Multiple nations
• Primary target: No clear primary
• Other affected: United States, Estonia, European Union
• Alleged perpetrator: Unknown

Threat characteristics

• Imminence: Long-term. This represents an ongoing commercial compliance and software quality issue rather than an active national security crisis.
• Autonomy: Full autonomy. The AI accessibility overlays operate and make real-time visual interpretation decisions on websites without active human intervention.
• Novelty: Established threat. Software defects, accessibility barriers, and regulatory non-compliance are well-established challenges in the digital ecosystem.

Impact by dimension

• Physical security: Negligible. The incident involves software failures in website accessibility overlays, posing no threat to physical security, critical infrastructure, or human safety.
• Information security: Negligible. No elements of information warfare, state-sponsored disinformation, or compromise of intelligence capabilities are present in this commercial software failure.
• Sovereignty: Negligible. There is no threat to state authority, territorial integrity, or core government operations. The impact is limited to private sector digital compliance.
• Economic security: Negligible. While companies faced legal fees and compliance lawsuits, this does not represent a threat to national economic stability, strategic industries, or technological competitive advantage.
• Societal stability: Minor. The incident highlights systemic failures in digital accessibility, leading to unequal treatment of disabled individuals and potential civil rights violations under the ADA. However, this poses minor threat to overall national security or societal stability.