Well, that's a whole area of active research, referred to as "info hazards":

https://www.lesswrong.com/w/information-hazards