Alibaba, ByteDance share algorithm details with Chinese regulator

609

BEIJING (AFP) – Chinese tech giants including Alibaba, Tencent and ByteDance have submitted details to a state regulator of algorithms used in some of their products, in an unprecedented move as authorities strengthen oversight of the sector.

The country’s Internet watchdog on Friday published a list of 30 algorithms that companies use to gather information on users and promote content or services.

Algorithms used by tech companies are closely guarded globally but in March China rolled out regulations obliging firms to disclose these tools as concerns grow about data misuse.

Beijing has in recent years embarked on a wide-ranging clampdown on the tech sector, which saw years of runaway growth and the emergence of supersized monopolies before regulators stepped in.

The rules issued in March require companies to ensure they will not engage in activities that may threaten national security, social stability or encourage over-indulgence, said associate professor of law at the University of Hong Kong Angela Zhang.

A man walks past the Tencent headquarters in Shenzhen in China’s southern Guangdong province. PHOTO: AFP

“It requires these service providers to conduct regular self-assessment to ensure compliance and to file their record with the relevant authority,” she told AFP.

Targets of the latest regulations range from China’s Twitter-like platform Weibo to delivery service Meituan and ByteDance’s short video app Douyin.

The March regulations say the rules apply to domestic service providers, meaning Douyin’s international cousin TikTok, which operates outside of China, would not have to share information.

The list published on Friday by the Cyberspace Administration of China briefly describes how the algorithms work and which products they are used for.

Alibaba’s popular e-commerce platform Taobao for example has an algorithm that recommends products based on users’ browsing and search history, while Douyin makes recommendations based on factors like how long a user engages with content.