This is a great question that I've talked to multiple people at Microsoft about. The way to ensure that legacy content is not used for Copilot is to ensure that legacy files are not accessible to search or deleted if they are not required for retention. If you can find a file with search, Copilot can consume it - for you, since you have access (so it's based on permissiosn). So, the bottom line is that you need to exclude content from search to ensure that Copilot does not consume the content. This is easier with files because you can ensure that they are in a library where search is disabled. It's harder with pages - so you need to make sure you shore up your governance plan! If the content isn't accurate, update it! If the content isn't needed, delete it! You can't currently use the term store to do this. You can, however, exclude entire sites from search. That will ensure that Copilot can't use the content. At the end of the day, it comes down to governance.
