Chinese Registrant Cited AI Tool DeepSeek in UDRP Defense, Lost LIBBS.com

The recent UDRP case involving the LIBBS.com domain name has attracted attention not only for the outcome but also for the registrant’s unusual defense: citing the Chinese AI search engine DeepSeek as evidence of non-infringement.

The domain was originally registered in 2008, deleted in 2022, and subsequently captured on DropCatch by a Chinese registrant named Li Lin Yang for $765. In early 2025, Brazilian pharmaceutical company Libbs Farmaceutica Ltda (which had owned the LIBBS trademark since 1961) filed a UDRP complaint seeking to reclaim the domain.

The respondent argued that “libbs” stood for “李家BBS”, a message board for the Li family, an acronym of his surname and the concept of “BBS”. To support his argument, he cited search results from the Chinese large language model and AI search engine DeepSeek. According to him, searching for "LIBBS" on DeepSeek did not turn up any information related to the pharmaceutical company, which he claimed proved that the term was not widely associated with the complainant.

However, the WIPO panel disagreed. They pointed out that the concept of "BBS" is outdated and questioned the feasibility of still using BBS in 2025. In addition, the defendant had no actual content on the domain and failed to provide credible evidence to support its claims.

The panel also pointed out inconsistencies in its alleged acquisition of the domain. The defendant claimed that he purchased LIBBS.com from a third party for $1,000, but WHOIS records showed that the buyer and seller shared the same phone number - suggesting that someone made up a backstory to legitimize the registration.

Ultimately, the panel ruled in favor of the complainant and ordered the transfer of LIBBS.com.

The case is not only noteworthy for its own controversy, but also because artificial intelligence tools like DeepSeek are beginning to make a name for themselves in legal debates. Despite DeepSeek’s innovative nature, the panel made clear that search results from AI models, no matter how advanced, have no legal validity in determining trademark rights.