The Cyberspace Administration of China (CAC) published rules banning the creation of deepfakes without the subject’s consent or to harm national interests. The Register.
According to the publication, the definition of national security encompasses anything that contradicts socialist values. Also, any form of “harmful information” or use of digital forgeries in an attempt to deceive or defame a person or group will also be illegal.
Nevertheless, the regulator expects more deepfakes in coming years, for example in chatbot contexts. In such scenarios, neural forgeries should be marked as digital creations.
According to the document, outlets that use deepfakes online must comply with many other Chinese rules governing acceptable content.
Creators of fakes, dubbed ‘deep-synthesis service providers’, must ensure the accuracy of artificial intelligence and machine-learning algorithms. The document obliges them to review regularly and ensure the security of the data collected.
The rules also include a requirement to register users, including their real names.
“[Deepfakes] should contribute to the healthy development of information services on the internet and support a healthy cyberspace ecology,” the document states.
The rules will take effect on January 10, 2023.
In January, the Chinese government introduced a bill to curb manufacture and dissemination of deepfakes.
In the same month, the CAC approved rules regulating recommender systems.
In November, the regulator announced a broad audit of algorithms for major websites, platforms and products that have a large influence on internet audiences.
Subscribe to ForkLog news on Telegram: ForkLog AI — all the news from the world of AI!