How to prove a legitimate interest in your .ai domain

A complaint lands. Someone with a trademark claims your .ai registration is abusive and demands a transfer. You know the name is yours by right – but knowing is not the same as proving. The question is what evidence the panel will actually credit, and how quickly you need to move.

To prove a legitimate interest in your .ai domain under a UDRP-based proceeding administered through WIPO, a registrant must bring their case within the safe harbors of Paragraph 4(c) of the UDRP: a bona fide offering of goods or services before notice of the dispute, being commonly known by the domain name, or legitimate noncommercial or fair use without intent to mislead. The respondent has 20 days to file a response once the case commences. A well-built record – use evidence, business documents, registration history – is what separates a successful defense from a default transfer.

This page covers the governing rules for .ai, the three Paragraph 4(c) safe harbors, how to assemble the evidence that matters, when a reverse domain name hijacking finding is realistic, and what the realistic next step looks like for a registrant ready to act.

Why the UDRP governs .ai disputes – and what that means for you

.ai is the country-code top-level domain for Anguilla, and its registry has adopted the UDRP as its dispute-resolution procedure, administered through WIPO. That means the same three-element test that applies to .com and .net applies to .ai: confusing similarity to a mark, no rights or legitimate interests in the registrant, and registration and use in bad faith – all three of which the complainant must prove.

The practical effect is significant. Because .ai has adopted the UDRP rather than a purely national procedure, the global body of UDRP panel decisions interpreting Paragraph 4(c) is directly applicable. A respondent defending a .ai domain can rely on the same established safe-harbor arguments that have succeeded in gTLD cases worldwide. The zone is familiar terrain to experienced respondent counsel, even if the registry is small.

What makes .ai unusual today is its commercial salience. Registrants acquired .ai domains long before the artificial intelligence sector exploded in market value. Many legitimate holders – technology companies, developers, researchers, domain investors with a credible AI thesis – now find themselves targeted by brand owners who registered trademarks recently and then filed complaints against older domains. That asymmetry matters when building a legitimate-interest defense.

One caution: because .ai is a ccTLD, specific registry eligibility rules may apply to the registration itself. Counsel should confirm the current registry requirements before filing any defense, because a technical ineligibility rarely defeats a respondent who otherwise satisfies Paragraph 4(c), but it is a point a well-prepared complainant will raise.

What are the three Paragraph 4(c) safe harbors, and which one fits your case?

Paragraph 4(c) of the UDRP provides three distinct grounds on which a respondent can demonstrate rights or legitimate interests, and any one of them is sufficient. The correct approach is to identify the strongest ground for your specific facts, then build the record around it rather than asserting all three weakly.

Safe harbor one – bona fide use before notice – requires evidence that the registrant was using or demonstrably preparing to use the domain in connection with a genuine offering of goods or services before receiving any notice of the dispute. Notice, in practice, typically arrives with the complaint or a cease-and-desist letter that preceded it. The key word is "demonstrably": contemporaneous records created before that date are far more persuasive than retrospective assertions. Development logs, invoices, wireframes, company formation documents, email threads, screenshots with metadata, and website hosting records dated to the pre-notice period all carry weight.

Safe harbor two – commonly known by the name – is most accessible to registrants who can show that the domain corresponds to their personal name, their business name, or a recognized abbreviation of either. This ground is frequently underused. A company whose legal name, trade name, or registered business name matches the domain, even loosely, can make a real argument here, provided the evidence of that recognition predates the complaint.

Safe harbor three – legitimate noncommercial or fair use – applies to commentary sites, fan pages, criticism domains, and genuinely noncommercial projects. It is harder to use where the domain is parked with pay-per-click advertising, because panels consistently treat PPC revenue as commercial use. If your .ai domain currently redirects to a parking page, that configuration is worth addressing before the response is filed.

If you have received a UDRP complaint targeting your .ai domain and need an immediate read on which safe harbor fits your facts, contact COGNOMEN at info@cognomenlaw.com. We assess the three UDRP elements, identify the safe harbor most supported by your evidence, and build the defense record before the 20-day response window closes.

How do you build the legitimate-interest record that wins?

Evidence quality decides .ai domain defense cases. A bare assertion – "I registered the domain for a legitimate project" – will not persuade a panel. The response must present a coherent narrative, supported by documents, that demonstrates the safe harbor is met.

The first step is a timeline audit. Map every material event: the date of domain registration, the date of any use or preparatory activity, the date of the complainant's trademark registration (and its claimed first use), and the date of the first communication alleging infringement. Panels examine this chronology carefully. A domain registered years before the trademark was filed, or before the complainant's brand reached any commercial significance, is strong evidence of good-faith registration. The burden of production sits with the complainant on elements one and three, but Paragraph 4(c) effectively shifts the narrative burden to the respondent once a prima facie case is raised.

The second step is document collection. Original registration confirmation emails, billing records, correspondence with developers or business partners, domain portfolio records, and any business registration documents linked to the name should all be compiled and date-verified. Screenshots are useful only when accompanied by metadata or independent corroboration – panels have seen fabricated screenshots, and unverified images carry limited weight.

The third step is addressing adverse facts directly. If the domain has, at any point, featured a parking page, a for-sale listing, or content targeting the complainant's brand, the response must explain that history rather than ignore it. Panels do not expect spotless records; they do expect honest ones. A respondent who acknowledges a period of inactive holding and explains it credibly – through illness, a project delay, or a pivot in business direction – is more persuasive than one whose response omits the fact entirely.

In a matter we handled in early 2025 involving a .ai domain in the technology sector, the registrant had held the name for several years before the complainant's trademark registration even issued. The response documented pre-dispute development activity, including dated repository commits and a vendor contract, and the complaint was denied. The panel found that the complainant had not met its own burden on registration in bad faith, and the respondent's legitimate-interest record was adequate to close the case independently.

When is an RDNH finding realistic, and how do you pursue one?

Reverse domain name hijacking – an RDNH finding – occurs when a UDRP panel concludes that the complaint was filed in bad faith, primarily to deprive a legitimate registrant of a domain they lawfully hold. RDNH carries no monetary penalty, but it is a formal finding in the published record and a meaningful reputational consequence for the complainant.

RDNH is realistic when the complainant demonstrably knew, or should have known, that it could not prevail. The clearest cases involve complainants who file despite clear pre-dispute evidence of the respondent's legitimate use, or whose trademark postdates the domain registration by a margin that makes bad faith on the respondent's part factually implausible. Panels have also found RDNH where a complainant filed with undue delay, where the complaint was factually thin, or where the real purpose appeared to be to recover a domain the complainant could not afford to buy on the open market.

To pursue an RDNH finding, the respondent must request it explicitly in the response. A panel will not make the finding sua sponte in most cases. The request should be structured: identify the specific deficiencies in the complaint, document the complainant's pre-dispute knowledge of your use, and demonstrate the procedural or substantive abuse. A request that simply asserts "this is RDNH" without analysis will rarely succeed; a request that walks through the chronology and the evidentiary record methodically gives the panel the basis to make a reasoned finding.

The value of an RDNH finding for a respondent extends beyond the immediate case. It creates a public record that may deter future abusive complaints, and it signals to the market – and to future panels – that the registrant is a serious holder, not a passive target. For domain investors who hold .ai names in a portfolio, an RDNH finding on one domain can provide useful precedent context in later disputes involving other names.

What evidence decides the outcome in .ai respondent cases?

Four categories of evidence most frequently determine whether a respondent succeeds or fails in a UDRP defense over a .ai domain.

First, the registration date relative to the trademark. If your registration predates the complainant's trademark filing, and certainly if it predates any evidence of use in commerce, the bad-faith element becomes very hard for the complainant to sustain. Document the registration date precisely, and obtain the complainant's trademark history from the relevant registry – it is public information and a necessary part of the analysis.

Second, documented pre-notice use. Use that predates the complaint, or documented preparatory activity that does the same, is the most reliable anchor for the Paragraph 4(c)(i) safe harbor. The earlier in the file history that use evidence appears, the stronger the defense.

Third, the absence of targeting conduct. Panels assess whether the registrant's conduct suggests the domain was chosen because of the complainant's mark. A domain acquired through a broad portfolio strategy, or because "AI" as a two-letter string was commercially generic before any single brand dominated the term, is less susceptible to a finding that the mark was targeted. Contemporaneous notes, acquisition records, and portfolio documentation all help demonstrate non-targeting.

Fourth, current use and configuration. How the domain is configured when the complaint is filed will be scrutinized. A domain that resolves to a developed website, a clear project page, or a professional company landing page is far more defensible than one pointing to a generic parking page. If the current configuration is weak, the response period is the time to address it – though any remediation must be authentic, not cosmetic.

In a second matter from summer 2025, a registrant holding a generic .ai domain was targeted by a complainant whose mark was registered only after the domain had been in continuous use for a technology newsletter. The complaint proceeded on a speculative bad-faith theory. Our response assembled the publication archive, subscription records, and the original domain registration confirmation. The panel denied the complaint, finding no credible basis for bad-faith registration given the timeline.

How does .ai compare to gTLD and other ccTLD procedures for respondents?

Because .ai has adopted the UDRP rather than a bespoke national procedure, the defense landscape is substantially the same as for .com. That is an advantage for experienced respondent counsel: the Paragraph 4(c) safe harbors, the burden-of-proof allocation, the RDNH doctrine, and the panel appointment mechanics are all familiar. The same panel pool that decides .com cases decides .ai cases at WIPO.

The contrast with some other ccTLDs is instructive. Nominet's DRS for .uk uses a different test – "abusive registration" – and reads the abuse element as "registered OR used" in an abusive way, a disjunctive standard that is procedurally distinct from the UDRP's cumulative "registered AND used in bad faith." A respondent who could argue good-faith registration under the UDRP's conjunctive test may face a harder path under the Nominet DRS if current use is problematic. The EURid procedure for .eu has its own eligibility requirements and a broader conception of protected "rights." For .de, there is no UDRP equivalent at all – disputes proceed through the German courts, with DENIC's DISPUTE entry available to block transfer during litigation.

For .ai respondents, the practical advantage is that WIPO is the forum, the procedure is the UDRP, and the jurisprudence is deep. A respondent with a legitimate claim to the name is not navigating an unfamiliar national procedure – they are working within a well-documented body of panel decisions where the standards are knowable and the arguments have been tested thousands of times. That said, the panel decides on the record before it, and every defense turns on its own facts. The body of prior decisions provides a map; it does not provide a guarantee.

If a prior response or filing produced an adverse result, a focused second read on the evidence record can identify what was missing. Email info@cognomenlaw.com to discuss whether a reconsideration path or a parallel action is available for your .ai domain.

What does the process look like from complaint to decision?

A UDRP case at WIPO – the most common forum for .ai disputes – moves through five stages: complaint filing and compliance review, case commencement and notification to the registrant, the response window, panel appointment and deliberation, and registrar implementation of the decision.

The response window is 20 days from commencement. That window is fixed. Extensions are rarely granted, and panels treat defaults as concessions – not as neutral non-participation. A registrant who does not respond gives the panel only the complainant's record to work from.

A standard case is typically decided within approximately two months of filing. Where a three-member panel is requested – either by the complainant or by the respondent – the timeline extends modestly, and the three-member fee applies. If the complainant requested a single panelist but the respondent requests a three-member panel, the parties generally split the higher fee. For a .ai defense where RDNH is a realistic argument or the facts are genuinely contested, requesting a three-member panel is often worth the additional cost: a collegial panel brings greater deliberative rigor, and an RDNH finding from three panelists carries more reputational weight than one from a sole panelist.

WIPO's filing fee for a single-member panel covering one to five domains is USD 1,500. That fee is paid by the complainant; respondents do not pay the forum fee. The cost of the defense falls on the respondent's legal representation. Market rates for respondent defense in a standard UDRP matter typically run in a range broadly comparable to complainant-side representation – fact-dependent, but usually within the USD 3,000 to USD 7,000 range for a single domain. Complex matters with a developed RDNH argument or contested factual record will run toward the higher end or beyond it.

What myths cost respondents their .ai domains?

Several persistent misconceptions lead registrants to underinvest in their defense, or to default entirely. Addressing them directly saves domains.

The first myth is that a domain registered years ago is automatically safe. Age of registration is significant evidence, but it is not conclusive on its own. A panel can find bad faith even in an old registration if the complainant produces credible evidence that the registrant knew of the mark at the time. Age is a factor, not a complete answer.

The second myth is that a generic or descriptive domain name is always defensible. "AI" is generic, and many two- or three-character .ai domains were acquired as generic technology identifiers. But if the specific domain string corresponds closely to a distinctive brand, and especially if the complainant can show the registrant had knowledge of that brand, the generic character of the extension alone will not carry the defense.

The third myth is that responding pro se – without legal representation – is adequate in a straightforward case. UDRP panels are not courts. They do not give a respondent credit for good intentions badly expressed. A response that fails to directly engage with each of the three UDRP elements, that omits a request for RDNH where the facts support it, or that presents evidence without proper context is measurably weaker than one prepared by experienced respondent counsel. The 20-day window is short enough that the cost of doing it twice does not arise: a default or a procedurally deficient response cannot be appealed in the ordinary course.

Related at COGNOMEN

• UDRP respondent defense and RDNH – the full service overview for registrants facing a complaint
• Defending an investment domain in the finance sector – strategy, safe harbors, and RDNH arguments
• Recovery after a failed buyback attempt for a .ai domain – the litigation and arbitration routes

Author: Anton Grant — Respondent defense and reverse domain name hijacking (RDNH). Meet the team.

Speak with Cognomen Law

For a scoped view of your domain matter, contact info@cognomenlaw.com. Discuss your matter

Domain in disputeYour sideI want to recover a nameI received a complaint — I need a defensePortfolio / brand protectionWhat happenedEmailI agree to the processing of my enquiry as described in the privacy policy.Assess my case

No obligation. Confidential. We reply within two hours in business hours.

This publication is general information and does not constitute legal advice. For advice on your situation, contact info@cognomenlaw.com.