Questions buyers ask
Chatbot accessibility and ADA Title II: common questions
Short answers for public colleges, school districts, cities and counties, each with its source. WCAG 2.1 AA is the standard in the ADA Title II rule.
Short answers
- Does ADA Title II apply to our chatbot?
- Very likely, for public entities: Title II covers web content, including a chatbot, that you provide.
- Which deadline applies to my college, school district, or city?
- April 26, 2027 for entities of 50,000 or more people; April 26, 2028 for smaller ones.
- Is WCAG 2.1 or 2.2 required?
- WCAG 2.1 AA is the Title II baseline; WCAG 2.2 is voluntary unless a contract requires it.
- Can our accessibility scanner test a chatbot?
- Partly: a scanner checks the page that hosts the chatbot, not the conversation in the chat window.
- Is a vendor VPAT enough?
- Not on its own: a VPAT is the vendor's self-description, not a test of your chatbot.
- What does a human reviewer do?
- A named reviewer uses the chat window by keyboard and screen reader and checks answers against official sources.
- Does student data leave our network?
- No logins or student records are needed: a reviewed check uses your public chatbot and pages.
- Does Title II cover a chatbot a vendor runs for us?
- Yes: the Title II rule covers web content you provide through a vendor, including its chatbot.
- Is the free scan enough on its own?
- No: the free scan is automated, covers only the page that hosts the chatbot, and has no reviewer.
- What is in the reviewed check report?
- A dated report: findings by severity, the reviewer's name and date on each, and vendor-neutral fixes.
- Does the report prove our chatbot meets the rule?
- No: it is dated evidence of what a reviewer found and your team approved, not a legal opinion.
- Is this an accessibility overlay?
- No: it is a human-led check, not an overlay, and nothing is added to your site.
- What about PDFs and other documents?
- Not yet: we do not check PDFs or documents today.
- How much does a reviewed check cost?
- A spot check is $950 per chatbot; a baseline with one retest is $2,450 per chatbot.
- How long does a reviewed check take?
- A spot check takes 3 business days and a baseline 10 business days, from the day we start.
- Can we recheck after a vendor or model update?
- Yes: annual assurance ($4,800 a year) includes a retest after each update, up to four.
- How do we buy?
- Email sales@syswisdom.ai; we send a PayPal invoice, then run the check.
- Can it run on our own hardware?
- Yes: the On-Site agent runs from a tar file on your own hardware, quoted from $2,000.
- Is the check based on MITRE's Chatbot Accessibility Playbook?
- Yes: the reviewed check is aligned with MITRE's playbook, used with the authors' permission.
Full answers, with sources
General information, not legal advice.
Does ADA Title II apply to our chatbot?
If you are a state or local government entity, such as a public college, school district, city or county, very likely yes. The Title II web rule covers the web content and mobile apps you provide or make available, including through contracts with vendors. A chatbot on your website is part of that web content.
Source: ADA.gov fact sheet on the web rule (opens in a new tab)
Which deadline applies to my college, school district, or city?
It depends on population. Entities with a population of 50,000 or more: April 26, 2027. Entities under 50,000 and special district governments: April 26, 2028. A city school district uses the city's population, a county school district uses the county's, and an independent school district uses its Census SAIPE estimate. A college follows the population of the public entity that governs it, not student enrollment; a college run by a special district government: April 26, 2028. Check with your counsel.
Source: ADA.gov fact sheet and the April 2026 extension (opens in a new tab)
Is WCAG 2.1 or 2.2 required?
WCAG 2.1 Level AA is the technical standard in the Title II rule. WCAG 2.2 is a voluntary best practice unless a contract requires it. Meeting 2.2 is a good goal, but 2.1 AA is the legal baseline.
Source: ADA.gov fact sheet on the web rule (opens in a new tab)
Can our accessibility scanner test a chatbot?
Partly. An automated scanner checks the page that hosts the chatbot, and only what has loaded when it runs. Many chat windows open only after a click, and no tool can judge whether an answer is right. The W3C notes that tools cannot check everything and human judgment is required.
Source: W3C: Selecting web accessibility evaluation tools (opens in a new tab)
Is a vendor VPAT enough?
A VPAT is the vendor's own description of how its product conforms. It is useful input, but it is not a test of the chatbot as configured on your site. The rule covers content you provide through vendors too, so test the chatbot where your users meet it and keep a dated record.
Source: ADA.gov fact sheet on the web rule (opens in a new tab)
What does a human reviewer do?
In a reviewed check, a named reviewer uses the chat window by keyboard only and with a screen reader, and checks a set of answers against your official sources. Each finding is written up, and a person on your team approves it before it goes in the dated record. The self-serve scan is automated only, covers the hosting page only, and has no reviewer.
Source: How the review works
Does student data leave our network?
A reviewed check uses your public chatbot and public pages, so we do not need logins or student records. With On-Site, the agent runs from a tar file on your own hardware with outbound network access blocked, which the free Proof Kit lets you verify, so the data you check stays inside your walls.
Source: Vault Proof Kit
Does Title II cover a chatbot a vendor runs for us?
The Title II web rule covers the web content and mobile apps a public entity provides or makes available, including through contracts with vendors. A chatbot on your site that a vendor runs is content you provide, so test it where your users meet it.
Source: ADA.gov fact sheet on the web rule (opens in a new tab)
Is the free scan enough on its own?
The self-serve scan runs automated rules on the page that hosts your chatbot and returns results right away. It does not open the chat window or judge answers, and automated checks cannot find every issue. A reviewed check adds a named reviewer.
Source: How the review works
What is in the reviewed check report?
Issues on the page that hosts your chatbot, grouped by severity. Findings from the reviewer's keyboard and screen reader pass inside the chat window. Answer accuracy findings checked against your official sources. Missing governance pieces, such as a vendor VPAT, an accessibility statement or a named owner. A person on your team approves each finding before it goes in the report.
Source: How the review works
Does the report prove our chatbot meets the rule?
SysWisdom does not certify anything and is not a legal service. The report is a dated record of what was tested, what was found and who approved each finding. Ask your counsel how to use it.
Source: How the review works
Is this an accessibility overlay?
Nothing is installed on your website and no script changes what your users see. A named reviewer tests your chatbot as it is, and you decide what to fix.
Source: How the review works
What about PDFs and other documents?
We do not check PDFs or documents yet. A reviewed check covers the page that hosts your chatbot and the chat window itself.
Source: How the review works
How much does a reviewed check cost?
The automated scan is free. A spot check is $950 per chatbot, a baseline with one retest is $2,450 per chatbot, and annual assurance is $4,800 per chatbot, per year. Chatbot vendors: $4,500 per product. The On-Site agent is quoted, from $2,000. To buy, email sales@syswisdom.ai and we send a PayPal invoice.
Source: Reviewed check prices
How long does a reviewed check take?
A spot check report arrives within 3 business days of the day the check starts. A baseline report arrives within 10 business days, and its retest is scheduled with you within 60 days.
Source: Reviewed check prices
Can we recheck after a vendor or model update?
A chatbot can change when its vendor or AI model changes. Annual assurance is $4,800 per chatbot, per year: the baseline, a retest after each vendor or model update (up to four a year), and a quarterly answer accuracy check.
Source: Reviewed check prices
How do we buy?
Email sales@syswisdom.ai about your chatbot. We send a PayPal invoice, then run the reviewed check or help you install the On-Site agent.
Source: Reviewed check prices
Can it run on our own hardware?
The On-Site agent ships as a tar file you run on your own hardware, with Vault and the Proof Kit, so your data stays inside your walls. Accessibility domain wisdom for On-Site is coming and is authored for you per engagement.
Source: Vault Proof Kit
Is the check based on MITRE's Chatbot Accessibility Playbook?
The reviewed check is aligned with MITRE's Chatbot Accessibility Playbook, used with the authors' permission. Findings are mapped to the playbook's recommendations and their WCAG success criteria. SysWisdom is not affiliated with or endorsed by MITRE.
Source: MITRE Chatbot Accessibility Playbook (opens in a new tab)
Get product updates
Get updates on the AI Quality Agent for accessibility. One email when there is news. No spam.