The Beijing Internet Court recently disclosed the trial of a related lawsuit: platform user Xiao Tang posted text content that they claimed was original, but the platform determined it "contained AI-generated content without labeling" and hid it and imposed a one-day mute. After an unsuccessful appeal, Xiao Tang sued in court. The court ultimately found that the platform had breached the contract and ordered it to lift the hiding of the content at issue and delete the corresponding backend records of the violation handling.
The case arose from a platform requirement that creators proactively label AI-generated content. Gao Ya, a judge in the Case Filing Division of the Beijing Internet Court, introduced that the defendant determined the answer at issue was AI-generated content through machine identification and manual review, and argued that it did not need to explain the determination mechanism to the user on the grounds that it involved technical secrets and algorithmic logic. The court held that the answer at issue was an improvised creative text, and that in litigation it would be difficult to require the plaintiff to provide drafts, original files, or other materials documenting the creation process; because the algorithm tool's computing process and review results were controlled by the defendant, the platform should reasonably provide evidence or an explanation for the basis of its determination. Since the defendant did not provide an adequate explanation, it should bear liability for breach of contract for handling the content without a factual basis.
The case's typical significance lies in clarifying the platform's duty to explain and the issue of remedies for misjudgment. The court pointed out that when a platform uses algorithms to identify AI-generated content and takes handling measures, it should explain the algorithm's operating mechanism and misjudgment remedy measures in a way users can understand, and cannot refuse to explain solely on the grounds of trade secrets. The court also suggested that platforms establish a tiered handling mechanism, adhere to the principle of proportionality between the violation and the penalty, adopt differentiated measures such as warnings, function restrictions, and service suspension according to the degree of violation, and improve content review mechanisms.
Against the backdrop of the implementation of the Measures for the Labeling of AI-Generated Synthetic Content and platforms' general strengthening of reviews of AI content labeling, this case provides a judicial reference for balancing algorithmic determinations and user rights. It also reminds platforms that while fulfilling their labeling management responsibilities, they must simultaneously improve explanations of the basis for determinations and channels for correcting misjudgments.
