Last updated on
Review Website Accessibility with AI and NVDA
Use an AI-assisted review to inspect a local game, hear NVDA feedback and turn the findings into specific manual accessibility checks.
Watch the video
Loading this video connects you to YouTube (Google), which receives your IP address and may store or read information on your device for playback and advertising. YouTube may show ads. Choose Load YouTube video to allow this for this video. See Google's privacy policy.
Watch Audit Website Accessibility with GPT-6 Astra and NVDA on YouTube
I ask GPT-6 Astra to inspect my local tic tac toe game while I follow along with NVDA on Windows 11. What I get back is a list of things to check myself: keyboard behavior, focus, and what the game says out loud.
Prepare a page you can test
Start the game on your computer with my run the game walkthrough. In the recording it is at http://localhost:5173. Make sure the game opens before you ask for the review. The Django API lesson covers the rest of the project.
I use Codex mode in the ChatGPT desktop app with GPT-6 Astra. Your model and menu names may be different. What matters is telling it the exact page, what to try, and what you want back.
Ask for both a description and interaction
My recorded request asks for an accessibility audit, a description of the page because I am blind, and an attempt to play the game. A more scoped follow-up prompt you can adapt is:
Review the game at http://localhost:5173. Describe its layout.
Try the keyboard controls, complete a game, and restart it.
Report what you tested, what you observed, and what remains untested.
Separate confirmed problems from suggestions and uncertain findings.
The address points at your own computer. When the app asked for access to the local site, I chose Allow once. Read what it is asking for before you answer a prompt like that.
Follow the game state with NVDA
Listen for each square's row and column, the mark you placed, whose turn is next, and the win and restart announcements. In my recording, X wins and I start a new game, and the tool also reports testing a draw.
Turn the report into checks you can repeat
The report describes a narrow layout at 320 pixels, contrast checks and suggestions about stable focus, reviewing the whole board and naming the winning line. For each finding, note the starting state, action, expected result and actual result. Then repeat the relevant interaction yourself with the keyboard and NVDA.
For example, after a win, verify where focus remains and whether you can understand the result without moving away from the board. After Play again, check that the new turn and available squares make sense. A clear spoken status and a predictable focus position solve different problems, so test both.
Know what the review did not cover
One correction: I say WCAG 2.1 near the end of the video, but the report uses WCAG 2.2. The report also lists checks it did not do: JAWS as well as NVDA, 200 percent text resizing, 400 percent browser zoom, text-spacing overrides, Windows high contrast and mobile screen readers.
So I treat the AI report as a starting point for my own testing, not a pass or fail for the site. The W3C evaluation overview explains why human testing still matters.
Save a useful follow-up record
Keep the date, page version, browser, screen reader, what you tested and what happened. Give each confirmed problem a small example you can repeat, and run it again after the fix. The video shows how I use AI to help me inspect a page, and the notes are what let me come back to it later.