It seems most of the "Hazards" outlined in the article are caused by the information causing the Doublethink machinery that maintains the receivers' social model to have to work harder. This isn't so much the information being harmful in of itself as the internally inconsistent models being harmed by factual information.