AI Disclosure & Transparency

ADAWCAG.org uses artificial intelligence in specific, limited places: to draft plain-English summaries of scan findings, to suggest remediation approaches, to answer product questions in our in-app assistant, and to handle initial intake on our inbound phone line. Every one of those uses is disclosed on this page, and each is labeled where you encounter it.

AI does not decide whether your website is accessible. Automated scanning identifies candidate issues; a DHS Trusted Tester–certified human auditor performs the manual testing, determines conformance, and reviews audit deliverables before they reach you. Where you see AI-generated text in a report, it is marked as such.

We publish this page because we sell compliance, and a compliance vendor that is vague about its own use of AI has no business auditing anyone else. If anything here is unclear or appears inaccurate, contact us at contact@adawcag.org and we will correct it.

The table below lists every feature or workflow where an AI model processes data on our platform. Vendor facts are derived from the sub-processor list in our Privacy Policy §5, which is the authoritative list of data-processing agreements. We will notify active customers at least 14 days before adding or replacing a sub-processor that handles personal data.

“Hi, this is Aria, an automated assistant for ADAWCAG.org. This call may be recorded for quality assurance. I’m here to help with accessibility compliance questions and to connect you with our team. How can I help you today?”

To be specific about the boundaries:

• AI does not perform the manual audit. Manual testing is conducted by a certified human auditor using assistive technology — screen readers, keyboard-only navigation, and related tools.
• AI does not determine conformance findings. Whether a success criterion is supported, partially supported, or not supported is a human determination.
• AI does not author or sign the VPAT/ACR. Accessibility Conformance Reports are prepared and attested to by a human.
• AI does not set or adjust pricing. Quotes are prepared by a person.
• AI does not approve, deny, suspend, or price any account. No eligibility, billing, or access decision is made automatically.
• AI does not deploy changes to your website. We do not push automated code changes to customer sites. Remediation, where you purchase it, is performed and reviewed by people.
• We do not use your data to train AI models, and our AI vendors are contractually prohibited from training on our inputs. See the sub-processor table in our Privacy Policy for the current list of vendors and the terms that apply to each.

We do not engage in “automated decision-making, including profiling” that produces legal or similarly significant effects on you under GDPR Art. 22. Our AI features generate human-reviewable summaries; they do not deny service, set pricing, or make eligibility decisions automatically.

We do not train AI models on your data, and we do not permit our vendors to do so. The vendors we use for AI features are contractually barred from using our inputs — including scan results, report content, and any user-provided text — to train, fine-tune, or improve their models.

• Anthropic: We use Anthropic’s zero data-retention API tier. Inputs are not logged or used for training.
• OpenAI: We operate under enterprise terms that prohibit training on our API inputs.

If this policy changes in any material way, we will notify active customers at least 14 days in advance before the change takes effect.

Automated scanning is incomplete by design. No automated tool can detect every accessibility barrier. Substantial categories of WCAG requirements — meaningful alternative text, logical reading and focus order, whether an error message is actually understandable, whether a custom widget behaves correctly with a screen reader — require human judgment. This is why our audits include manual testing, and it is why we say plainly that automated tools alone are not sufficient for compliance.

AI-generated text can contain errors. Summaries and remediation suggestions produced by AI are drafted from scan output and may be incomplete, may misstate the cause of an issue, or may propose a fix that does not suit your codebase. Treat them as a starting point for your developer, not as a specification. Where an AI-generated section appears in a deliverable, it is labeled, and it has been reviewed by an auditor before delivery.

A scan is not a legal finding. Running a scan, receiving a report, or purchasing monitoring does not guarantee compliance with the ADA, Section 508, WCAG, the European Accessibility Act, or any other legal requirement, and does not guarantee that you will not receive a demand letter or be sued. Accessibility is an ongoing obligation; a site that conforms today can regress with the next content update.

We are not a law firm and do not provide legal advice. Nothing on this site, in a scan report, in an audit deliverable, or in a conversation with our AI assistant or our staff constitutes legal advice, and no attorney-client relationship is created. For advice about your specific legal exposure, consult a qualified attorney.

Tell us when we get it wrong. If an AI-generated section of a report is inaccurate, email contact@adawcag.org and request a human re-review. We will re-examine the finding at no charge.

You have the right to:

• Request human review of any AI-generated section of your accessibility report. Email contact@adawcag.org with the subject line “Request for human review of AI report section” and include your scan URL or report ID. We will respond within 5 business days.
• Speak with a human at any point in the Aria phone intake flow by saying “agent” or “human.” Aria will immediately transfer your call.
• Opt out of AI-assisted features where technically feasible. Contact contact@adawcag.org to discuss available options for your account.
• Exercise your privacy rights (access, correction, deletion, portability) as described in our Privacy Policy §8.

For any AI-related concern not covered above, email contact@adawcag.org or call 844-232-9224.

This page is published in light of the transparency requirements below. We describe our own practices here. We do not claim certification, approval, accreditation, or endorsement by any regulator, agency, or standards body in connection with our use of AI, and none of the frameworks listed below confers such a status.

• EU AI Act, Article 50 (transparency obligations, applicable from 2 August 2026). Where a person interacts with one of our AI systems, we inform them at the point of interaction. Our in-app assistant is labeled as AI before you send a first message; our inbound phone intake identifies itself as automated at the start of the call; AI-generated sections of reports are labeled in the deliverable.
• California SB 1001 (the “B.O.T. Act,” Bus. & Prof. Code §17940 et seq.). Our automated systems identify themselves as automated in communications with users. We do not use a bot to mislead anyone about its artificial identity.
• California SB 243 (companion chatbots, effective 1 January 2026). Our assistants are customer-service and operational tools. They are not companion chatbots and are not designed to meet a user’s social or emotional needs. We disclose their automated nature regardless.
• California CCPA/CPRA and other U.S. state privacy laws. What we collect, how long we keep it, who processes it, and how to exercise your rights are set out in our Privacy Policy. We honor the Global Privacy Control signal.
• Section 5 of the FTC Act (unfair or deceptive acts or practices). We do not overstate what our AI features can do. Where automation has limits, we state them — see Accuracy and Limitations above.
• GDPR Article 22. We do not engage in automated decision-making, including profiling, that produces legal or similarly significant effects on individuals. See our Privacy Policy for the full statement.

This list is not exhaustive, and the regulatory landscape for AI is changing quickly. We review this page when our AI usage changes and when applicable requirements change; the effective date above reflects the most recent revision.

Effective: August 17, 2026
Last updated: August 17, 2026

Material changes to this disclosure will be announced by email to active customers at least 14 days in advance. A full history of changes to our product and policies is available at adawcag.org/changelog.