In the rapidly evolving landscape of artificial intelligence, the concept of ‘AI safety’ is often framed around protecting human values and well-being. However, a critical dimension is frequently overlooked: safeguarding society from authoritarian abuses. Recent reports highlight how regimes like the Chinese Communist Party are leveraging AI technologies to enhance surveillance and control, raising urgent questions about the implications for democracies worldwide. As AI systems become more integrated into governance and law enforcement, the potential for misuse grows exponentially.