China’s cyberspace authority listed a total of 41 deep synthesis service algorithms developed by Chinese internet companies on Tuesday. Observers said the move will help draw up the legal framework for generative artificial intelligence (AI), a highly competitive high-tech sector which is experiencing fast-lane development in China.
The 41 items on the list include algorithms for trending AI innovations including big data, graphic generation and identification, video generation, voice generation and identification, intelligent customers service and face fusion, underscoring Chinese technology companies’ breakthroughs in different AI fields.
The technologies are developed by Chinese tech giants such as Baidu, Tencent, Alibaba, ByteDance, iFlyTek and Meitu, the Global Times has learned.
The technology breakthroughs specify the corresponding application scenarios for individual algorithms. For example, the voice identification technology developed by iFlyTek is applicable in voice generation scenarios, under which the developer is allowed to extract the voice characters from voice record, and combine it with algorithm to identify short voice clip and generate text information.
Another example is the facial fusion algorithm developed by Shenzhen-based Bighead Bros which owns video app Doupai. The algorithm could fuse the image users uploaded with specific human image to generate facial image and video that shows the facial characters of the users.
Deep synthesis technology involves deep learning, machine learning and other algorithmic processing system, and is commonly known as “deepfake” in technical terms. It uses mixed datasets and AI algorithms to produce synthetic yet “fake” content. The technology has quickly gained vogue in China, after the worldwide popularity of ChatGPT – an AI-generated chatbot which is also deemed as an advanced application of deepfake technology.
“It is a timely response to new developments in the AI market. The application of which algorithm is legitimate? Who could use it? In which cases? The record clarifies some vague points and is conducive to the regulated growth of this significant yet double-edged technology,” Wang Peng, a research fellow from the Beijing Academy of Social Sciences, told the Global Times on Wednesday.
Amid white-hot global competition and the US-led blockade against China’s tech rise, the legal framework provides a foundation based upon which Chinese tech companies are assured to apply big data model to improve and perfect the AI algorithms, offering a boost for relevant technologies’ broad applications, Wang noted.
Regulating how algorithms are applied is also helpful for strengthening data security, observers said.
The file, adding to a number of regulations issued in recent months on AI-generated content, have jointly set a fence preventing the potential abuse of deepfake technology in China, which has already brought challenges to supervision such as copyright infringement, law-breaking practices such as fraud, and even harms the national security and social stability.
The publication of the 41 algorithms is in accordance with the management regulation on deep synthesis of internet information service, which went effective in January 2023. Under the regulation, deep synthesis service providers with public opinion attributes or social mobilization capabilities shall register their service. Any changes and cancellations of their deep synthesis service should go through set procedures.
(Global Times)
