Meta Oversight Board finds top AI models less likely to criticize repressive regimes - Reuters

Top AI models made by major companies are less likely to criticize authoritarian governments, which means they may help spread propaganda without pushback. This affects what information billions of people see when using these tools.