China’s Latest Move Targets AI Content with Strict Labelling Rules – Will It Work?

China Tightens Control Over AI Content with New Regulations

China’s internet regulator is intensifying its grip on AI-generated content through a newly proposed regime that mandates all digital platforms label AI-created material.

Announced by the Cyberspace Administration of China, this draft plan seeks to enforce a transparent labelling system across various forms of content, from text and images to video and audio files.

If adopted, platforms and online service providers must ensure AI-generated material is easily identifiable through visible logos and embedded metadata.

The regulations will go even further for audio files, where a voice prompt must inform listeners at the start and end of a file – and sometimes mid-file – that they are engaging with AI-generated content.

Similarly, software playing these files will also need to notify users upon interaction with AI material.

For videos, Beijing will allow platforms to display notices at the beginning, end, and critical points in the content.

An anonymous lawyer specialising in intellectual property transactions, based in Beijing, noted:

"This is about protecting the integrity of the digital ecosystem,"

Tightening Surveillance on User-Generated AI Content

The draft also extends to individuals posting AI-generated content.

Netizens will be obligated to label any content they generate using AI, and those utilising platform tools will be required to identify themselves, leaving a digital footprint.

Logs of user activities must be stored for a minimum of six months to ensure accountability.

This measure marks yet another step towards China’s broader objective of linking all user-generated content to identifiable individuals, reinforcing the country’s ambition for sovereign control over its cyberspace.

The implications are clear – content that challenges the Communist Party or the government is unlikely to survive in China’s highly monitored internet environment.

Zhang Linghan, a professor at the China University of Political Science and Law who supports the labelling system's potential to improve AI-generated content governance, cautioned:

"Failing to meet these stringent requirements could land both platforms and individuals in hot water."

He emphasised that the regulations would reduce misinformation, prevent the corruption of future AI training datasets, and create a healthier digital ecosystem.

Analysing AI-Generated Content When Metadata is Missing

To ensure compliance, service providers must also analyse content lacking metadata.

If a provider suspects that AI has generated a file, it will have to label it accordingly.

This aspect of the draft seeks to eliminate any loopholes in the system and ensure that all AI-generated content is identifiable, even if metadata is deliberately omitted.

Additionally, the regulations call for platforms to share the technologies they develop to detect AI content, fostering a collective effort to improve detection methods across the board.

These measures come as part of China’s broader strategy to secure national interests and safeguard public security.

Building on Earlier Regulations to Address Misinformation

This new draft regulation builds on a series of earlier frameworks designed to regulate the use of artificial intelligence.

One key milestone was the Administrative Provisions on Deep Synthesis in Internet-based Information Services, implemented in January 2023.

This policy already required conspicuous labelling of content that might mislead or confuse the public, paving the way for the more stringent guidelines proposed in the current draft.

Another crucial development occurred in August 2023 with the Interim Measures for the Management of Generative AI Services, China’s first set of rules specifically aimed at AI-generated services.

This ongoing regulatory progression highlights China’s proactive approach to managing the rapid advancements of AI technology.

China is not the only country grappling with the implications of AI-driven content.

Other governments and organisations worldwide are implementing similar measures.

For example, the European Union’s Coordinated Plan on Artificial Intelligence, the US Deepfakes Accountability Act, and the UK’s Online Harms White Paper all reflect global efforts to ensure the authenticity of online content and protect intellectual property.

AI and Deepfake Technology Under Scrutiny

The rise of deepfake technology, which relies on AI to manipulate images, audio, and video, has heightened concerns about the spread of misinformation and potential security risks.

In a recent interview with the People’s Daily, Professor Zhang emphasised the importance of the new labelling system, stating:

"The labelling system will enhance governance capabilities for AI-generated content, reduce the production of misinformation, prevent the contamination of future training datasets, and foster a healthy cyber information ecosystem."

Deepfakes have already been exploited for fraudulent purposes, further reinforcing the necessity for stringent regulations.

In this context, China’s proposed laws aim to mitigate the risks associated with these technologies, emphasising the importance of protecting national security and maintaining public trust.

By labelling AI-generated content and holding platforms accountable, China aims to create a more transparent digital space.

A Push Towards High Compliance Standards in AI Governance

The draft regulations not only seek to control AI-generated content but also signal the beginning of a more intense era of regulatory compliance in China’s generative AI sector.

A Beijing-based lawyer specialising in intellectual property observed,

"This initiative marks a new phase of stringent regulation and high compliance standards in China’s generative AI services sector."

By imposing higher requirements on platforms, the Chinese government ensures that AI technologies are used responsibly and that the negative impacts of AI-generated content are minimised.

As public consultation on this draft opens for the next month, stakeholders will have an opportunity to provide feedback.

However, based on historical precedent, Beijing rarely makes significant alterations after unveiling drafts like this.

It is likely that the proposed rules will be implemented in their current form, further tightening control over China’s digital landscape while protecting the country's interests in the face of rapidly evolving AI technologies.