This is plausibly bad if this behavior shows up in future catastrophes that kill eg. billions of people.

It's bad if this behavior shows up in future catastrophes IFF different behavior was available (knowable and achievable in terms of coordination) that would have reduced or mitigated the disaster. I argue that the world is fragile enough today that different behavior is not achievable far enough in advance of the currently-believable catastrophes to make much of a difference.

If you can't do anything effective, you may well be better off optimizing happiness experienced both before the disaster occurs and in the potential universes where the disaster doesn't occur.