AI Content Labels Backfire? Study Finds Real Info Trust Drops, Fake Info Trust Rises

AI-generated scientific content on social media is increasing, raising concerns about misinformation. Many countries are pushing for AI content labeling to protect the public. However, a new study in the Journal of Science Communication shows such labels may backfire. A team from the University of Chinese Academy of Social Sciences found AI labels reduce trust in real content but increase trust in fake content. Using GPT-4 to create real and fake posts, 433 participants' ratings revealed this 'truth-fake cross effect' clearly. Researchers suggest double labeling (e.g., adding 'not independently verified' statements) or risk-based classification to avoid the contradictory effects of single labels.