> ...wouldn't that still imply that Google are parsing Docs in some manner if they're able to be flagged for ToS violations in general?

Public documents in Google Docs are used as targets for spam or phishing campaigns pretty frequently. (The latter is usually by way of Google Forms, which can be used to drop content into a spreadsheet.) Google needs some way to allow this behavior to be flagged, and ideally to recognize it before it gets abused.

Private documents are another matter. Even there, though, I imagine they are obliged to scan some content, e.g. checking images against child pornography databases.

is there empirical evidence of this type of pornography check being performed?

No, that may be a generic message wrongly assigned to the observed behavior.

I get it now. It may have (falsely) determined a violation from their behavior/use of Docs instead of the actual content of those documents.

Nevermind my original comment then. That makes sense.

Well, the fact that Google scans your documents as you're writing them, and that they can summarily remove your access to the thing that you were working on for an indefinite amount of time, are both pretty scary for someone using Google Docs for work.

I'm sure it's just fine most of the time but I don't understand how people trust it for mission critical (or even just important) stuff.