UserAgent: Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/63.0.3239.132 Safari/537.36

Steps to reproduce the problem:
1. Perform an action that causes a web dialog to spawn, an example would be deleting a comment in the Chromium tracker.
2. When the dialog displays, NVDA says "unknown."  
3. Press tab, visual focus moves through the dialog options, but nothing is read.
4. Attempt to explore the dialog using nVDA's object navigation commands, this also will fail.  In summary, NVDA appears to not be aware of the contents of the dialog at all.
5. Note that pressing enter on an option will execute the option as expected, so this issue seems very specific to NVDA and not to keyboard interaction.

I have not tested this with other screen readers at this time, I'll comment if i am able to duplicate this on ChromeOS or anywhere else.

What is the expected behavior?
When I tab through the options within a dialog and they gain focus, NVDA should read the options.  Also, if i disable NVDa's "browse" mode, I should be able to review the contents (including static text) of the dialog.  Lastly, I should be able to read the dialog using NVDA's "object nav" functionality as this should enable me to read through all visible objects on-screen.

What went wrong?
NVDA says "unknown" and is unable to read the contents of the dialog, including static text and options.

Did this work before? No 

Chrome version: 63.0.3239.132  Channel: stable
OS Version: 10.0
Flash Version: