BIC’s inaugural 'search suggestion algorithm' infringement case takes effect
On Nov 11, 2024, the Beijing Internet Court (BIC) released its judgment in the first instance trial of the court's inaugural online infringement case involving "search suggestion words".
The court held that the search suggestions at issue were automatically generated and updated by the defendant's algorithm based on the historical records of searches and browsing by unspecified users. The defendant did not manually add new content or specifically aggregate negative content, nor did they conduct manual pre-audits.
Given the vast number of users, the high search volume and the instantaneous, dynamic, and massive nature of search suggestion words, it is not technically feasible for the defendant to pre-audit every search suggestion for potential infringement within reasonable costs. During the trial, the plaintiff questioned whether the defendant selectively added search suggestion terms to certain disputed infringement content, arguing that this constituted human intervention. The plaintiff contended that this subjective intervention served to amplify the scope of infringing content dissemination and infringement impact using the algorithm. The plaintiff further requested the defendant to clarify the algorithm rules and generation principles of the involved suggestion words.
In response, the defendant has twice provided written explanations to the BIC on the generation mechanism of the search suggestion technical service, and the basic principles, operational rules and technical feasibility of displaying the suggestion words.
The defendant's explanations effectively addressed the algorithmic risks posed by the search suggestions in question, their causes, and the possibility of avoidance, hence fulfilling the burden of proof and can be regarded having fulfilled related explanatory obligations.
Accordingly, the BIC ruled that the defendant's provision of the search suggestion service did not constitute an infringement on the plaintiff's right to reputation. The verdict has now taken effect.
The principles of algorithm explanation and transparency are key elements of algorithm governance systems both domestically and internationally. These principles have been reflected in China's institutional framework and practices, though the development of judicial rules based on specific application scenarios still requires improvement. For instance, when legal entities are involved as the subjects of algorithm applications, do they have the right to request algorithm explanations? In addition to individual information processors, should the developers and designers of the algorithm, as well as its users, also be obliged to provide algorithm explanations? What should be the form and content of algorithm explanations? When trade secrets are involved to, do the standards for pre- and post-explanation differ, and what legal liabilities arise from a breach of the obligation to explain the algorithm?
In this case, it recognizes that the defendant platform's search suggestion algorithm service involves infringement. And the plaintiff, as the party affected by the algorithm application, has the right to request an explanation of the algorithm. As a user of the algorithm, the defendant should also adhere to the principles of fairness, openness and integrity, and provide timely and reasonable explanations regarding the basic operation of the algorithm and the basis of the algorithm results. These explanations constitute post hoc explanations of the damage that has already occurred. The defendant cannot refuse to explain solely on the grounds of trade secrets. Otherwise, the defendant may be legally liable in reference to the burden of proof and the elements of fault.
The case was widely picked up by mainstream media outlets, including Guangming Online, Legal Daily, Workers Daily and Beijing Daily, after its ruling was announced.