Beijing has unveiled a draft law to regulate how Chinese tech companies use algorithms, which are at the heart of nearly every profitable tech business today, from news aggregation and short videos to ride-hailing, e-commerce, and food delivery. Manish Singh wrote a summary of the policy, and here’s a detailed look at China’s top cyberspace watchdog’s 30-point proposal.
Beijing is plainly concerned about how entirely machine-recommended content can deviate from Communist Party norms, even to the harm of national interests. Algorithms, in its opinion, should rigorously correspond with the national interest: Mainstream values should be upheld by algorithmic suggestions and should not be utilized in any way that jeopardizes national security. Regulators want firms’ algorithmic black boxes to be more transparent, and they’re holding them liable for the outcomes of their computer codes. For instance, service providers should be accountable for algorithm security. They should set up a system for reviewing published material, algorithmic procedures, and security oversight, establish and publish algorithmic recommendation rules that are appropriate. Users should not be enticed into an addiction, high-value consumption, or any other behavior that violates public order by service providers.
In addition, the government is cracking down on discriminatory algorithms and giving customers more control: User interests should not be based on illegal or damaging information, nor should sexist or prejudiced user tags be created.
Users should be informed about the logic, intent, and methods of algorithms in use by service providers. Users should be able to disable algorithmic features from service providers. Regulators do not want internet powerhouses to sway public opinion or thinking. Even though it is not stated in the text, censorship control will almost certainly stay in the authorities’ control.
Service providers should not employ algorithms to block information, make massive recommendations, manipulate ranks or search results to provide themselves preferential treatment or unfair competition, sway online opinions, or avoid regulatory inspection. Specific algorithms, like many other elements of the computer industry, must be approved by the government. In the event of an investigation, IT companies must also send over their algorithms to the police. If service providers’ recommendation algorithms have the potential to influence public opinion or mobilize civilians, they should file a report with the government.
Service providers shall preserve a record of their suggestion algorithms for at least six months and offer it to law administration departments for investigation reasons. If passed, the regulation will upend Chinese internet companies’ fundamental business models, which depend on algorithms to generate money. Programmers must be able to decipher these rules and interpret their code for regulators. The new law appears to go beyond the limits of data protection standards in the European Union, but how the Chinese one will be implemented remains to be seen.
Xiaomi said on its most recent earnings call that it will purchase DeepMotion, a Beijing-centered autonomous driving business, to help with its autonomous driving efforts. The deal would cost Xiaomi $77.3 million, with “a lot of that in terms of stock” and “a lot of such payments deferred until specific milestones are met,” according to Xiaomi president Wang Xiang, who spoke on the call. Xiaomi’s founder, Lei Jun, had previously hinted at the company’s intention to enter the congested market. Lei revealed on Weibo, China’s counterpart of Twitter, on July 28 that the company is hiring 500 autonomous driving professionals across the country.