The report found that users curious about dubious posts were directed to more false claims and dangerous content.