China cracks down on use of live-streaming and AI to sell religion (www.ft.com)

Chinese regulators have announced a clampdown on using live‑streaming and AI tools to promote or commercialize religion, ordering platforms and content creators to stop monetized religious broadcasts, sales of religious items via live commerce, and the production or distribution of AI‑generated religious content (for example synthetic sermons, cloned voices, or avatar preachers). Platforms will be required to detect and remove offending content, tighten account verification and payment flows, and face penalties for noncompliance. The move frames AI‑enabled synthetic media and algorithmic recommendation of religious material as a regulatory risk and extends existing controls on online religious activity into the era of generative models and live e‑commerce. For the AI/ML community this raises immediate technical and governance implications: teams must improve real‑time detection of synthetic audio/video and identify religious content with low latency for live moderation; integrate provenance, watermarking, and robust classifiers while managing false positives and adversarial evasion; and adapt recommendation systems to avoid surfacing disallowed material. Platform engineers will need to balance automated enforcement with transparency and appeals processes to mitigate overblocking. More broadly, the decision underscores how public policy is shaping acceptable uses of generative AI, creating region‑specific compliance demands for model training, data curation, and deployment practices for companies operating in or serving users in China.