AI Porn Content Chain Exposed Amid Crackdown

Artificial intelligence is being used to generate and profit from pornographic content through a growing underground industry chain, China Central Television reported yesterday.

An investigation by CCTV's Financial Investigation program found that vendors across multiple online platforms openly promote guides on how to create pornographic videos using AI. These posts, often paired with provocative imagery, claim users can generate explicit content from a single image or a few lines of text, and even earn substantial daily income.

Sellers also offer so-called "prompt packages" for less than 10 yuan (about $1.4), which can be directly used in AI video-generation tools. These prompts include highly detailed instructions covering appearance, clothing, poses and movements, while deliberately using coded or translated language to bypass content moderation systems.

Tests conducted by CCTV reporters on 10 widely used AI applications showed that although most platforms block overtly explicit requests, rephrasing prompts in more subtle ways can allow users to evade safeguards. In one case, an app identified as "LiblibAI" generated a semi-nude dancing video within minutes without triggering any restrictions.

The investigation also found that some businesses profit by distributing overseas AI chat applications featuring explicit interactions. These apps, not available through official domestic channels, are circulated via private links, with promoters offering incentives such as virtual credits to recruit new users.

In one such app, "Dady," users could create customized AI companions engaging in sexually suggestive conversations, with responses tailored to user input. The immersive experience has attracted a growing number of users, including minors, raising concerns among experts.

Regulators in China have stepped up enforcement. The Cyberspace Administration of China has launched a campaign targeting the misuse of AI technologies, with AI-generated pornography listed as a key focus.

New measures on the management of anthropomorphic AI interaction services, released on April 10, impose stricter requirements, including safety assessments and algorithm registration.