China tells AI to stop emoting online; cracks down on malicious pessimism, ennui (www.theregister.com)

China’s Cyberspace Administration launched a two-month campaign to stamp out online content that “maliciously incites negative emotions,” ordering social media, short-video, livestream commerce platforms and comment sections to remove material that promotes violence, coordinated harassment, conspiracy theories, or “excessively exaggerated negative and pessimistic sentiment” — examples cited include slogans from “Sang culture” such as “hard work is useless.” The directive explicitly covers AI-generated content when it depicts violence, and threatens platforms and creators with punishment and forced “rectification” for violations. It also singles out tactics like leveraging fan communities to organize mass abuse or complaints. For the AI/ML community this raises practical and policy implications: platforms will need to scale more nuanced moderation that distinguishes criminal incitement from subjective pessimism, and to detect AI-generated media, coordinated campaigns, and sentiment patterns in real time. That intensifies demands on automated classifiers, content provenance tools, and moderation pipelines — increasing false-positive risk and creating legal/compliance pressure on model deployment and training data curation. Because Beijing has repeatedly issued similar crackdowns, the move highlights ongoing tension between algorithmic recommendation loops that amplify user sentiment and government efforts to control online mood, with potential chilling effects on creative or critical AI-produced content and research practices.