CHINA Trends and Developments Contributed by: Dr James Luo and Angie Guo, Lawjay Partners
prompt design, systematic parameter adjustment, or curated output selection. This “human authorship” standard, however, remains confined to copyright law and does not govern trademark registration. Trademark framework: focus on distinctiveness Under Articles 8 and 9 of China’s Trademark Law , a sign’s registrability is determined primarily by its distinctiveness and capacity to identify commer- cial origin. The China National Intellectual Property Administration (CNIPA) Trademark Examination and Trial Guidelines further specify that eligibility depends exclusively on a sign’s capability to distinguish goods or services – regardless of its creative process. Notably, China’s trademark regime imposes no requirement for human authorship. Legal protection extends to the sign itself as a source identifier, rather than its creative process. Therefore, an AI-generated design may be registered as a trademark if it possess- es the required distinctiveness, provided the applicant qualifies as a “natural person, legal person, or other organisation” under Article 4 of the Trademark Law . However, in examination practice, explicitly disclos- ing a mark as “AI-generated” may trigger heightened scrutiny. Examiners could question the sign’s inherent distinctiveness or potentially invoke Article 10 (1)(8) of the Trademark Law concerning marks deemed to have “adverse effects on public order or social morality.” Third parties may use such disclosures to challenge (Article 33) or invalidate (Article 44), especially in cases involving well-known trademarks. To mitigate registration risks, applicants should adopt prudent disclosure strategies: • avoid terminology implying fully automated AI generation; • characterise the mark as “the final design devel- oped by creative professionals using AI-assisted tools”; and • maintain thorough documentation of human crea- tive direction (including prompt iterations, output selection records, manual refinements, internal approval processes, etc).
This evidentiary foundation, while not a statutory requirement, serves as a strategic safeguard – a defensive reserve to address potential challenges during examination or post-registration proceedings. AI-Generated Infringement: Redefining Likelihood Traditional trademark infringement analysis under Arti- cle 57 (2) of China’s Trademark Law has historically relied on assessing visual, phonetic, and conceptual similarity to determine the likelihood of confusion. However, generative AI has introduced a sophisti- cated new paradigm: algorithmic pattern replication that systematically recreates the distinctive essence of protected marks while avoiding literal copying. For example, when prompted to “generate a pattern similar to LV Monogram,” AI tools typically produce outputs that replicate the core visual architecture – including distinctive composition principles, colour relationships, and textural patterns – through para- metric reconstruction. This process creates designs that are “form-variant but spirit-similar,” preserving the original mark’s essential commercial impression and distinctive characteristics while avoiding pixel-level duplication. Legal Framework: beyond literal comparison Although AI-generated content may not technical- ly infringe copyright, it raises important trademark issues related to benefiting from established brand reputation and undermining the source-identification function. The critical legal question becomes whether these algorithmically derived signs create a sufficient likelihood of confusion to constitute infringement under Article 57 (2) of the Trademark Law . of Confusion in the Algorithmic Age The challenge of “intelligent imitation” The Hangzhou Internet Court’s landmark 2023 judg- ment established crucial guidance on this matter, determining that: • trademark similarity assessment must transcend the mechanical comparison of individual elements; • analysis should focus on “overall visual effect” and “degree of inheritance of distinctive features”; and
209 CHAMBERS.COM
Powered by FlippingBook