[dredmorbius]

The problem with pre-generated activity is that it must cut off at (or shortly in the future of) generation time, to be plausible.

Otherwise that activity would be suspicious due to either a lack of recent records, or of presumably implausible future ones.

Generating data in advance and applying or updating timestamps later, on an ongoing basis, or when a duress code is entered is a possible way of mitigating this. There's the question of how convincing such data would have to be. White-flagging and generating (or appropriating from public sources, e.g., business or institutional entities) contacts for this might be a part of it. This is similar to but not entirely the same as data fuzzing, which is generally seen as applying to a primary data trail.