With new regulation in effect, local tech companies should expect greater scrutiny over how they harness users’ data to promote content and services
Algorithm information is typically not made public: The Cyberspace Administration of China (CAC) has published descriptions of 30 algorithms of domestic tech firms, including Alibaba, Baidu and Tencent. The inner workings of these algorithms, which decide what users see and the order they see it in, are usually a closely-guarded secret. The CAC’s unprecedented move follows March 2022 regulation, which gives consumers the right to be informed and/or opt out of personalised recommendations driven by algorithms – and also requires companies to ensure that their algorithms promote “positive energy”. The regulator’s 30-strong list includes Weibo’s “personalised push” algorithm, which is used in content recommendations and is based on historic browsing behaviour. Similarly, ByteDance-owned Douyin has a recommendation algorithm for images, videos, goods and services that leverages users' past clicks, likes, comments and shares.
China has sought to curb the power and influence of the tech sector: Publishing algorithm details may be unsettling for the companies involved, although the descriptions themselves are fairly superficial and do not include lines of code. As disclosing such intellectual property could harm prized local tech firms, the CAC’s action appears to be about control and utilising new powers under the March regulation. China’s huge – and growing – internet user base has enabled the rapid expansion of local device, e-commerce, gaming and social media companies. However, in recent years, the Government has tightened its grip on these firms in light of concerns over the potential misuse of data and the influence they could have on shaping public opinion online, particularly as platforms for sharing false information. In July 2022, the CAC fined Didi $1.2bn for breaching data security rules, while the Government has also stepped in to limit the time children can spend gaming each week, which has contributed to Tencent’s falling revenue growth.
US firms have argued that their algorithms represent business secrets: It remains to be seen whether the CAC’s transparency drive will trigger similar requests from policymakers elsewhere. Neither the EU or the US has introduced legislation around proprietary recommendation algorithms, with China the only country to reveal the mechanisms that underpin purchasing and content viewing decisions. In the US, Alphabet and Meta have resisted pressure to submit this kind of information, previously arguing (successfully) that their algorithms amount to trade secrets. It is also unclear what confidential information Alibaba, ByteDance and their peers may have shared with the CAC privately, and what the regulator’s next move might be. Nevertheless, releasing the list of 30 algorithms indicates that implementing the March regulation is progressing as intended. Though the CAC has stated that it wants to support the “healthy development” of domestic tech companies, industry should brace itself for greater scrutiny of algorithms and for further public disclosures.
Source: http://www.cac.gov.cn/2022-08/12/c_1661927474338504.htm
