Watchdog Finds AI Chatbots Take Sides on Political Speech

Meta's Oversight Board tested 10 AI chatbots from OpenAI, Google, Anthropic, xAI, and Meta. When asked about political protest, the chatbots tended to discourage criticism of governments that already restrict speech, while encouraging criticism of governments that allow it. Co-chair Paolo Carozza called the pattern "extended censorship by proxy that goes across borders."