AI technology was originally an important force driving social progress, but when it is transformed into a tool that breaks the legal bottom line and violates public order and good customs, it becomes a "weapon" that harms society.

On April 12th, CCTV Finance's "Financial Investigation" program heavily exposed that a complete black industry chain of AI "pornography" and the dissemination of obscene and pornographic content was rampant on the internet. Criminals openly sell 9.9 yuan AI pornography tutorials and pornographic prompts. With just one photo or a few paragraphs of text, they can generate realistic pornographic borderline videos in bulk. They also teach users how to avoid regulatory measures, forming a complete closed loop of "tutorials prompts tools monetization".

After this incident was exposed, the whole internet was in an uproar. A netizen angrily criticized, 'It's so unethical to use AI technology in this place for 9.9 yuan to destroy a person!'! ?Some parents are also concerned: "Nowadays, children can see borderline content everywhere online, and AI 'pornography' will seriously harm minors. ?Industry insiders also bluntly stated, "Technology is innocent, but those who misuse technology for evil must be severely punished, and platform supervision cannot be absent

This illegal industry chain hidden in the digital economy is using technology as a weapon to constantly challenge the legal bottom line and public order and good customs, turning AI, this cutting-edge technology, into a tool for illegal crimes and becoming a social cancer that urgently needs to be eradicated.

AI's' yellow making 'operation is covert

On multiple Internet platforms, an illegal profit making chain around AI technology is spreading openly. Various tutorial posts, under the banner of "AI made beauty videos" and "AI borderline track entering five digits daily", openly promote how to create pornographic borderline content through AI technology "from scratch", packaging illegal activities as a so-called "entrepreneurial trend" and openly spreading them in cyberspace.

CCTV reporters found through investigation that one of the core links in the industry chain is the low-priced sale of so-called "prompt words" - these key instructions that command AI work directly determine the nature of the generated content. The reporter successfully purchased two sets of prompts from stores such as "Wanxiang Xiaodian" and "Nannan Material Library" at low prices of only 9.9 yuan and 9.8 yuan. Upon opening the file, it is evident that the prompt words meticulously depict details such as hairstyles, clothing combinations, and even waist movements of the characters. In order to evade supervision, the merchants deliberately translate sensitive vocabulary into English, forming a hidden command system.

The operation of this illegal industry chain has formed a clear division of labor system, presenting a complete closed loop of "tutorial sales - prompt word sales - software generation - content dissemination". In order to make a profit, these businesses not only teach production techniques, but also teach step-by-step tricks to avoid regulation: first translate sensitive words into English, then replace explicit instructions with seemingly normal descriptions such as "hairstyle, waist and hip twisting amplitude", use overseas TuSheng video AI software, deploy programs offline on local computers, and other means to completely bypass platform regulation.

For this reason, CCTV reporters conducted actual tests on 10 AI applications that users use more frequently, and found that if the prompt words entered are too explicit, most AI applications will trigger the review mechanism and refuse to generate content. However, as long as the prompt words purchased from merchants that have undergone "technical processing" are used, it is possible to easily bypass the restrictions.

In an app called "Libulibu A", after the reporter entered the purchase prompt words, within just a few minutes, the platform generated a semi naked female dance video without triggering any restriction mechanism throughout the process. This indicates that although major platforms have safety barriers, there are still obvious blind spots in the existing review technology when faced with carefully designed "adversarial prompts".

It is worth noting that the tentacles of this illegal industry chain also extend to the field of "virtual companionship". Some overseas AI chat apps cannot be downloaded normally through domestic app stores, but on social media platforms, coded posts with "can drive, unlimited words" can be seen everywhere, and download links and invitation codes are secretly spread among users.

In addition to exploiting vulnerabilities in legitimate AI platforms, illegal software that specifically relies on AI to generate pornographic videos for profit has also emerged in the market. According to consumer reports, CCTV reporters have found a software that claims to be download free and can quickly "change faces". As long as the buyer is willing to pay, the software can generate nude photos and pornographic videos from a single image. These types of software are often spread through covert channels, with clear pricing, such as "10 yuan per image, 1 yuan per second for video", and their generated content is highly realistic, almost difficult to distinguish between real and fake.

Starting point of industrial chain

According to CCTV reports, AI "pornography" illegal and criminal activities are rampant, and criminals almost do everything they can to make a profit. The industrial chain has formed a clear division of labor and a certain scale of "upstream, midstream, and downstream", and the operation mode is characterized by specialization, concealment, and maturity.

In the upstream of the industrial chain, criminals obtain a large number of facial photos and body data of ordinary people, especially women, through illegal channels. These photos may come from personal social media accounts that have been hacked, life photos that are publicly available but maliciously collected online, or even images that have been forged and synthesized.

Midstream technical operators use open-source or easily modified AI face swapping and image generation techniques such as deepfakes to replace the obtained normal portraits with the faces of characters in pornographic videos, or directly generate extremely realistic fake pornographic images and videos through text descriptions.

Recently, there has been a concentrated outbreak of abuse of AI "face swapping" technology, with individuals ranging from entertainment and sports stars to politicians, businessmen, and even ordinary people at risk of becoming victims. Recently, Hongguo Short Drama released a "Statement on Handling Violations of Platform Governance Norms", announcing the complete removal of the short drama "Peach Blossom Hairpin" that has attracted public attention due to AI infringement. Earlier, some netizens accused the drama of using its personal portrait without authorization and using AI technology to generate villainous characters, suspected of "face theft" infringement.

When the portrait of ordinary people may become the next target of abuse, AI is recklessly changing faces and becoming an accomplice to rampant digital bullying and "pornography". According to incomplete statistics, in the first three months of this year alone, dozens of actors and film and television professionals have publicly defended their rights against AI short dramas, and there are countless cases of portrait infringement against ordinary people.

AI face swapping was originally an innovative technology that could have become a typical example of technology for good. However, the reality is that it has gradually become a tool for "digital skinning" in a negative atmosphere of chasing traffic and illegal interests.

In the downstream dissemination and profit-making process, criminals sell and spread through various encrypted communication software, niche social platforms, pornographic websites, and even some communities disguised as "AI painting communication" both domestically and internationally, forming various profit models including customization, wholesale, membership based paid viewing, etc.

What is even more worrying is that these yellow contents often gain wider dissemination through precise algorithm recommendations, forming a vicious cycle that is difficult to break.

Governance: Strengthen legal deterrence and improve regulatory system

Addressing the chaos of "AI induced pornography" requires multi-party collaboration and comprehensive management.

Firstly, the technology platform should take on the main responsibility, strengthen content review, use AI technology to identify illegal content generated by AI, and establish a convenient channel for infringement complaints and handling. Secondly, developers of AI tools and models should also strengthen ethical design, embed necessary usage protocols and technical protections when open sourcing or providing services, and prevent models from being easily abused.

For the public, improving digital literacy and security awareness is crucial. It is important to be cautious when sharing photos containing clear facial and body features online, and to learn how to identify common vulnerabilities in deepfake content. The most important thing is that the judicial authorities need to increase their crackdown on this new type of cybercrime, form a deterrent, and continuously improve the judicial appraisal standards and evidence collection norms for technologies such as deep forgery.

Currently, China's legal supervision in related fields is rapidly following up. The Civil Code specifies the protection of portrait and reputation rights. The newly revised Cybersecurity Law, Data Security Law, and Guiding Opinions on Investigating and Dealing with Illegal and Criminal Activities Using AI Generated Content provide legal basis for cracking down on such behavior. There are clear charges in the Criminal Law to punish the act of producing and disseminating obscene materials for profit. However, facing the challenges of rapid technological iteration, innovative criminal methods, and difficulties in cross-border evidence collection, regulation and law enforcement still need to be continuously upgraded.

Previously, the Cyberspace Administration of China (Cyberspace Administration of China) accurately deployed the "Clear and Clean up AI Technology Abuse" special action, clearly identifying the use of AI to produce and publish pornographic and vulgar content as a key target for rectification, and comprehensively consolidating the platform's main responsibility. In April 2026, the Cyberspace Administration of China and five other departments jointly issued the "Interim Measures for the Management of Artificial Intelligence Personification Interactive Services", which will be implemented from July 15, 2026. This measure sets a red line for the application of AI technology at the institutional level, covering key areas such as algorithm filing and protection of minors, providing a solid institutional guarantee for regulatory work.

The Ministry of Industry and Information Technology and ten other departments recently jointly issued the "Measures for Ethical Review and Service of Artificial Intelligence Technology (Trial)", proposing to support technological innovation in ethical review of artificial intelligence technology and strengthen the prevention of ethical risks in artificial intelligence technology through technological means.

Despite continued tightening of regulations, the game between technology and rules continues. Among the multiple AI applications named by CCTV, there are many top products with over 100 million users. These platforms are not without review mechanisms, but when faced with implicit prompts such as "describing the waist curve as an S-shaped curve", the system has difficulty recognizing the implied meaning or has regulatory loopholes. This technological loophole and outdated rules keep criminals with a sense of luck, constantly trying to break through regulations through shortcuts and forming a continuous confrontation with them.

Only by continuously strengthening legal deterrence, improving regulatory systems, consolidating platform responsibilities, and enhancing technological recognition capabilities can AI technology return to the right path, cut off the survival space of illegal industrial chains, and safeguard a clear and healthy network ecology and social environment.