By Steve Levy
In all the years I’ve been filing UDRP complaints, there had only been a couple of cases where I’ve encountered evidence that was blatantly manufactured in an effort to show that a domain name owner had legitimate rights to the disputed domain name. In one case, a respondent submitted a modified corporate registration certificate from Malaysia and, in another case, fake posters were produced claiming that the respondent operated a legitimate business. In both cases, an adept and suspicious eye and careful sleuthing uncovered these fakes and, through opposing evidence, the UDRP panelists saw through these cybersquatters’ deceptions.
Recently, however, I encountered falsified evidence on an entirely different level. In this case, the respondent redirected the disputed domain name to the complainant’s own website and offered to sell it to the complainant for sums ranging from US $50,000 to $10.5 Million. This is fairly straightforward evidence of bad faith under par. 4(a)(iii) of the UDRP. However, in an effort to claim that he had rights or legitimate interests, under par. 4(a)(ii), the respondent produced a range of documents that I suspect were created through generative AI. First up was an agreement by which the respondent claimed that, as a 17 year old intern, he was granted a 2% equity stake in a global, multi-billion dollar company in exchange for creating a website using the disputed domain name. The language of the agreement was rather amateurish but the smoking gun was the use of a graphic signature of the company’s then CEO from a readily available public online source. In its decision, the UDRP panel noted that “Such an agreement is utterly fantastical and an obvious forgery.”
Next, was a single page from what was claimed to be an article in a major technology magazine describing the respondent’s work at the complainant’s company and displaying an image of the then CEO with his arm around the respondent. When pressed for further details, the respondent stated that he did not have the first or any other pages and did not know the title or the date of the article. Online searches of the publication’s database found no evidence of the article nor any mention of the respondent.
There was other evidence, such as a mocked-up intern badge and screenshots of a claimed US $80 Million third-party offer to buy the domain name. But the biggest smoking gun was a forged US trademark registration certificate. The respondent boldly claims to have owned a registration of the complainant’s trademark and he submitted a copy of the certificate. At first glance, it looked legitimate. However, when the registration was searched at the USPTO database, the result showed that it is for a completely different trademark. When pressed on this point, the respondent claimed that his registration had lapsed and the USPTO must have re-used the same registration number. This is obviously not the practice of the USPTO but the respondent claimed to have an email from someone at the office stating that they do, in fact, sometimes re-use registration numbers. However, the email was from a @gmail.com address and not the usual uspto.gov address used by the trademark office.
In the end, the panel found this to be “a series of clumsy and obvious forgeries…” and noted that “the evidence submitted by the Respondent in support of his claim is replete with falsities…” This raises the question of whether generative AI was used to produce some of these forgeries. As no sophisticated tools were used to identify the fraud, the situation may be old win in new bottles with the solution being a sharp eye and dogged detective work to prove the fabrications. Of course, the UDRP does not have strong evidentiary tools, such as discovery, depositions, witness cross-examination, etc. and so there may be future disputes that are not well suited for resolution through the policy. In those cases, brand owners may need to access the courts and hire expert witnesses, unfortunately, at far greater expense than a simple UDRP case. The one thing that seems certain, however, is that cybersquatters will continue trying to game the system in ever more sophisticated ways and brand owners, and their representatives, will need to be ready to spot and investigate evidence that appears too good to be true.