China proposes plan to require labeling of all AI-generated content posted online

The Cyberspace Administration of China (CAC), the country's internet regulator, has announced plans to require digital platforms to label all AI-generated content.

National Communication Security Information Office 《Artificial Intelligence Generation and Synthesis Contents 标识办法》Public release of information information Central Office

https://www.cac.gov.cn/2024-09/14/c_1728000676244628.htm

China wants red flags on all AI-generated content • The Register
https://www.theregister.com/2024/09/16/china_ai_content_draft_regulations/

China proposes mandatory red flags placed on all AI-generated content posted online | Tom's Hardware
https://www.tomshardware.com/tech-industry/artificial-intelligence/china-proposes-mandatory-red-flags-placed-on-all-ai-generated-content-posted-online

The Cyberspace Administration's plan would require platforms and online service providers to label all AI-generated content, including both 'explicit identifiers, such as logos or voice prompts that are clearly recognizable to users,' and 'non-user-visible identifiers that are added to the data in the content files.'

If the AI-generated content is text, it must be labeled by adding a written explanation or symbol at the beginning, end, or center of the text. In addition, in the case of audio files, it is necessary to provide a voice explanation to the listener at the beginning, end, or middle of the audio. Images and videos are also required to be labeled in a prominent position, so that users can understand whether the content was generated by AI.

Creators will be required to use the platform's labeling features if they post AI-generated content on the platform or use any AI-generated tools provided by the platform.

In addition, even if the posted content file does not contain metadata indicating that it was AI-generated, if the platform's analysis of the content finds evidence that it was generated by AI, the platform is required to notify users that 'this content is suspected to be generated by AI.'

This time, the Cyberspace Administration said it would 'seek public comment on the draft.' However, foreign media outlet The Register said, 'Chinese regulators are unlikely to make any major changes after such a draft is released. It is likely that this draft is something the Chinese government wants to implement,' suggesting that the draft may be enacted as is.

Technology media Tom's Hardware said that this draft is a typical example of the Chinese government's Internet regulation, but it's not all bad. In recent years, developers and vocal supporters of generative AI tools have tried to use them to replace skilled workers and artists, but if labeling of AI-generated content becomes mandatory, that could change the trend.

Legislation regarding generative AI is being developed not only in China but also in other countries around the world, and in the US state of California, a bill called 'AB 3211' that would require the labeling of AI-generated content is being deliberated. This bill is also supported by lobbying groups including OpenAI, Microsoft, Adobe, and Amazon Web Services (AWS).

OpenAI, Amazon, Microsoft and other major IT companies support bill to require labeling of AI-generated content - GIGAZINE