China is actively taking steps to increase transparency in information generated by artificial ielligence and published on the Iernet. The Cyberspace Administration of China (CAC), the coury’s national iernet regulatory body, recely released a draft law that includes guidelines for labeling AI-generated coe.
The law, titled “Artificial Ielligence-Generated Composite Coe Labeling Actions,” addresses providers of AI-generated text, images, audio, and video, and aims to increase transparency. This draft was inspired by legislation such as the Cyber Security Act and the AI Service Manageme Regulations.
The purpose of this law is to iroduce a single standard for monitoring coe related to artificial ielligence and reduce the amou of false information and deepfakes (deep fake coe) on the Iernet. According to this law, labels are divided io the following two categories:
Specific tags:
Visible marks such as warnings or watermarks should be placed on AI-generated text, images, audio, and video. For example, videos produced with artificial ielligence should have clear signs in their opening frame and warnings should be displayed in appropriate places in the text.
hidden tags:
Hidden data such as metadata or watermarks should be embedded in AI-generated files. These tokens include information such as coe source, AI service provider, and unique ideifier. These tags are not immediately visible, but can be recognized by platforms and regulatory bodies to verify coe autheicity.
The implemeation of these regulations is associated with obstacles. Among the most importa obstacles for the implemeation of such laws is the issue of financial costs related to its implemeation. Currely, platforms such as Xiaohongshu, Bilibili, Weibo, Douyin and Kuaishou, which are active in China, have implemeed rules related to coe generated by artificial ielligence, but its implemeation It will be difficult for smaller companies. Also, this action is an attempt to ensure national and public security in China.
Beijing has always been wary of emerging technologies and has been a step ahead of couries like the US or the EU in regulating them. Zhuang Rongwen, director of the Cyberspace Administration of China (CAC), is an influeial person not only in China but also in the West. He was recely named as one of the most influeial people in the Times 100 list. He has always tried to realize the values of the Chinese Communist Party (CCP) and has actively emphasized the coury’s power in the growing artificial ielligence (GenAI) competition with the West.
When ChatGPT launched in 2022, China iroduced a law the following year to preve the rapid and uncorolled growth of artificial ielligence. According to this law, companies must first obtain governme approval for public use of models.
Last year, Senator Mark Warner gave an ierview to Politico and said: “China is far ahead in the field of artificial ielligence self-regulation within its own coury,” and added: “China is the leader in artificial ielligence laws.” and has left other couries behind.
It was earlier this year that India’s Ministry of Information and Communication Technology (MeiTy) issued guidelines on labeling of AI-related coe online. This guide was published after problems with the Gemini AI and was later revised.
In a previous version of the guide, companies were told they would have to get governme approval before launching new AI models. This was met with a lot of criticism, as many believed that the move could limit innovation and slow down the process of developing new technologies.
