VPAT automation in 2026 is AI-driven, mapping audit data to WCAG success criteria and generating conformance levels with remarks. The critical factor is what data feeds the AI: scan results produce unreliable ACRs, while VPAT automation that is based on an accessibility audit will lead to strong automated population.
This is why it’s so important to use an audit-based platform.
| Approach | What It Means for ACR Quality |
|---|---|
| Scan-Based Automation | Uses automated scan results as the data source. Misses 50-70% of accessibility issues. Produces incomplete, unreliable ACRs that informed buyers will reject. |
| Audit-Based Automation | Uses findings from manual accessibility audits. Captures issues that only human testing can identify. Produces ACRs grounded in actual WCAG conformance evaluation. |
| Hybrid Model | AI handles initial population and remarks generation. Human expert reviews and corrects. Combines efficiency with accuracy verification. |
| Full Manual Process | Expert manually transfers audit results to VPAT template. Time-intensive but historically the only reliable method before AI capabilities matured. |
Foundation
Any AI-generated VPAT automation is only as good as its data source.
The Voluntary Product Accessibility Template (VPAT) documents how a product or service conforms to accessibility standards like WCAG. When filled in and completed, it becomes an Accessibility Conformance Report (ACR). The accuracy of that ACR depends entirely on what information feeds the AI.
Automated scanners catch certain issues: missing alt text, some color contrast failures, detectable code problems. They cannot evaluate whether a custom dropdown works with a screen reader, whether keyboard focus moves logically through a form, or whether error messages are announced to assistive technology users. These gaps matter because procurement agents rely on ACRs to make purchasing decisions.
When AI pulls from scan data alone, the resulting ACR reflects only partial testing. The conformance levels may look correct in the template, but they do not represent actual WCAG conformance. This creates liability for the vendor and frustration for buyers who discover accessibility barriers after purchase.
Real Automation
Proper VPAT automation starts with a manual accessibility audit.
An auditor tests the product using screen readers like NVDA, JAWS, or VoiceOver. They navigate with keyboard only. They inspect the code for proper ARIA implementation. They check visual presentation, text resizing, and reflow behavior. This evaluation produces a detailed report of accessibility issues mapped to specific WCAG success criteria.
That audit report becomes the data source for the AI. The software analyzes which success criteria have outstanding issues and which have been resolved. For criteria with validated fixes, the conformance level is “Supports.” For criteria with unresolved issues, the level is “Does Not Support” or “Partially Supports” depending on scope.
The remarks and explanations column is where AI saves the most time. Writing specific remarks for each of the 50+ WCAG 2.1 AA success criteria takes hours manually. AI generates these remarks from the issue descriptions in the audit report, producing draft language in seconds rather than hours.
But this is draft output. The human review layer remains essential.
Hybrid Automation
AI initially fills in the VPAT. A human expert reviews and corrects.
This hybrid model recognizes that AI makes errors. It may misclassify a conformance level. It may generate remarks that need refinement. The AI provides a starting point that dramatically reduces manual work, but it does not eliminate the need for expert oversight.
Accessible.org clients using Accessibility Tracker follow this workflow. The platform accepts audit reports, tracks remediation progress, and uses AI to generate VPAT documentation based on current issue status. The output is a draft that the user reviews, edits, and finalizes before it becomes an official ACR.
This differs fundamentally from tools that promise fully automated ACRs with no human verification. Those promises should raise immediate concerns about accuracy.
Automation Claims
When a vendor claims to offer VPAT automation, ask specific questions.
What is the data source? If the answer involves only automated scanning, the ACR will be incomplete. Manual audit findings should form the foundation. Scans can supplement but not replace human evaluation.
Is there a review step? Legitimate AI automation produces drafts for human verification. Fully automated output without expert review creates risk. Errors in an ACR can have legal implications and damage vendor credibility with buyers.
Who conducts the underlying audit? The technical accessibility experts conducting the evaluation should use diverse manual evaluation methodologies: screen reader testing, keyboard testing, visual inspection, code inspection. If the offer doesn’t start with an audit, the AI is building on a faulty foundation.
What VPAT editions are supported? Currently Accessibility Tracker supports automating the WCAG edition with Section 508, EU, and INT scheduled for Q2 of 2026.
Saving Time
The real efficiency gain is in the transfer and documentation phase.
Conducting the accessibility audit itself cannot be automated without sacrificing accuracy. That work requires trained professionals spending hours testing products and services against WCAG criteria. No shortcut exists for this foundational step.
What AI eliminates is the tedious process of moving audit findings into the VPAT template. Manually, this involves reviewing each success criterion, determining the conformance level based on related issues, and writing remarks that explain the assessment. For a comprehensive audit with dozens of issues across multiple screens, this documentation work alone can take several hours.
With audit-based AI automation, that transfer happens in minutes. The AI maps issues to criteria, generates conformance levels, and drafts remarks. The expert then reviews and refines rather than building from nothing.
Organizations producing multiple ACRs see compounding benefits. An accessibility company issuing reports to clients reduces time per project. An enterprise managing ACRs for a portfolio of products can scale documentation without proportionally scaling staff. Procurement timelines accelerate when documentation is not the bottleneck.
Audit-Based
AI changes how ACRs are assembled, not what makes them credible.
Buyers evaluating ACRs look for evidence of thorough evaluation. They check whether the remarks column contains specific details about the product rather than generic language. They assess whether the conformance levels seem consistent with the type of product being documented. They consider whether the ACR was issued by a reputable third party.
An ACR generated from scan data alone will fail this scrutiny regardless of how sophisticated the AI or how professionally the document is formatted. An ACR generated from comprehensive audit data will demonstrate the depth of evaluation that procurement agents expect.
The value proposition of AI-powered VPAT software is efficiency in documentation, not replacement of expertise. Tools that position themselves correctly deliver genuine time savings. Tools that promise to eliminate the audit altogether deliver unreliable documentation that creates problems downstream.
FAQ
Can AI fully automate VPAT creation?
AI can automate the documentation phase when working from manual audit data. It cannot automate the accessibility audit itself. Full automation claims that skip manual testing produce incomplete ACRs missing issues that only human evaluation can identify.
What makes audit-based automation different from scan-based automation?
Audit-based automation feeds AI with findings from manual testing using screen readers, keyboards, and code inspection. Scan-based automation feeds AI with automated tool output that misses the majority of accessibility issues. The data source determines ACR accuracy.
How much time does AI-powered VPAT automation save?
The documentation phase that previously took hours can be reduced to minutes. The audit itself still requires the same expert time. Organizations producing multiple ACRs see the largest cumulative time savings.
Should I trust a fully automated ACR Software?
Services claiming fully automated ACRs without manual auditing should be approached with caution. Accurate ACRs require human evaluation of WCAG conformance. AI works best in a hybrid model with expert review of its output.
What VPAT editions can be automated?
The WCAG edition is most commonly supported by AI automation tools. Section 508, EU, and INT editions follow similar logic and may be available depending on the platform. Check which editions a tool supports before committing.
