Introduction
The growth of artificial intelligence (AI) applications and brands is one of the most significant Internet trends of the last few years[1]. As for many other high-profile online phenomena, the popularity of AI can be seen 'mirrored' in trends in the domain-name landscape, with AI-related domain names being used to drive Internet traffic and promote a wide range of related products and services. We also find that high-profile AI names such as the OpenAI GPT / ChatGPT brands (found to be the most prominent AI brand by a significant margin, in a recent study[2]) specifically find themselves subject to a range of infringements, including instances where the brand name is used in domain names to promote third-party offerings and imply affiliation.
In this study, we consider the landscape of domain names related to AI and the GPT brand specifically.
Methodology
The dataset used in this study consists of domain names with names ending with 'ai' or 'gpt' (to reflect the manner in which the terms are frequently used, and to avoid the issue of non-relevant false positives which would arise by searching for any domains containing the (very common sub-)string 'ai'), focusing on gTLD (generic top-level domain) extensions for which zone-file data is available (as of 05-Dec-2024).
In total, there are over 828,000 domains ending with 'ai' (approx. 782,000, of which approx. 39,000 end with '-ai' (with an explicit hyphen) or 'ai' immediately preceded by a non-Latin character, i.e. those least likely to be 'false positives') or 'gpt' (approx. 47,000). Nearly 7,400 of these domains also include the keyword 'chat' (as a case study of those most likely to be making reference in some way to the ChatGPT brand and/or a common use-case (chatbots) for AI applications - and noting also that the chat.com domain had recently been acquired by OpenAI for a price in excess of $15M[3]).
In order to carry out a deeper dive into the data, we consider the subset of domains containing both 'chat' and (at the end) 'gpt', of which there were 4,274 gTLD examples as of the date of analysis.
Analysis
The first point to note is that none of the domains in the dataset (where whois information is available via an automated look-up) were registered prior to November 2022 (the month of launch of the 'research preview' of ChatGPT[4]). However, immediately following this date, there was a very large spike in registration activity (Figure 1).
Figure 1: Numbers of 'chat' + 'gpt' gTLD domains, by month of registration
Secondly, only 27 of the domains (0.6%) appear to be under the official ownership of ChatGPT brand owner OpenAI, with the remainder controlled by third parties and constituting potential brand infringements. Statistics for the set of third-party domains are shown in Tables 1 and 2.
| TLD |
No. domains |
|---|---|
| com | 2,487 |
| chat | 295 |
| net | 166 |
| org | 124 |
| top | 108 |
| online | 93 |
| xyz | 78 |
| info | 56 |
| site | 40 |
| shop | 37 |
| store | 36 |
| pro | 30 |
| vip | 22 |
| app | 22 |
| tech | 21 |
Table 1: Top TLDs for the third-party 'chat' + 'gpt' domains
| Registrar |
No. domains |
|---|---|
| GoDaddy.com, LLC | 1,087 |
| Alibaba Cloud Computing Ltd. d/b/a HiChina (www.net.cn) | 384 |
| NameCheap, Inc. | 378 |
| Squarespace Domains LLC | 251 |
| DNSPod, Inc. | 113 |
| NameSilo, LLC | 108 |
| Dynadot Inc | 102 |
| Porkbun LLC | 88 |
| IONOS SE | 64 |
| Amazon Registrar, Inc. | 58 |
Table 2: Top registrars for the third-party 'chat' + 'gpt' domains
The prevalence of new-gTLD extensions which have previously been noted as having high incidence rates of fraudulent use[5,6], and of retail-grade registrars (also typically popular with infringers), within the dataset is striking.
1,194 (28%) of the third-party domains produce some sort of live website response. Many of these resolve to neutral or low-threat content (e.g. informational or placeholder pages) but there are numerous examples which appear potentially to be infringing the ChatGPT brand. Some examples are shown in Figure 2.
- Log-in page (potential phishing) (examples: 91chatgpt[.]fun, lanchatgpt[.]work, chat2gpt[.]vip, yourchatgpt[.]org)
- Other chatbot application (examples: xachatgpt[.]top, cw-chatgpt[.]com, 4-chatgpt[.]com, chatsgpt[.]pro, btm-chatgpt[.]asia and six other similar sites; all generate browser warnings of dangerous content, upgpt[.]chat)
- Misdirection of web traffic / promotion of competitors (example: ok-chatgpt[.]com)
- Claimed official / affiliated site (examples: chat-gpt[.]bond and five other similar sites, anonchatgpt[.]com)
- e-commerce site (examples: 51chatgpt[.]info, 58chatgpt[.]com)
Next, for those domains containing 'chatgpt' as a full exact string, it is informative to consider the portion of the second-level domain name (SLD, i.e. the part of the domain name to the left of the dot) prior to this string, as an indication of the keyword patterns used in the set of registered domains. These strings include a wide range of generic terms, plus a number of other brand names (Figure 3) (potentially also comprising infringements against these brands) - and suggest that 'ChatGPT' has come some way towards being used as a generic term for AI applications.
Figure 3: Wordcloud of most common terms appearing prior to 'chatgpt' in the domain names within the dataset (courtesy of https://www.voc.ai/tools/wordcloud - powered by ChatGPT(!))
The additional brands featured in these domain names include both AI-related and non-AI examples, with the list including Apple, Bard, Bentley, Boeing, eBay, Gemini, Google, Grok, Huawei, iPhone, Meta, Midjourney, Rolex, Spotify, Tiktok, Tinder, Uber and Zoom. The majority of these do not resolve to content any more significant than parking or placeholder pages (though do indicate some intention to monetise the domains, through the inclusion of pay-per-click links or offers of sale for the domain names), but some live additional potential infringements were identified (Figure 4).
Figure 4: Examples of potentially infringing websites hosted on domains featuring ChatGPT and another brand name -
- (top) potential misdirection / unauthorised claim of affiliation (names: googlechatgpt[.]mobi, googlechatgpt[.]space, googlechatgpt[.]work)
- (bottom) potential unauthorised use of brand name (name: zoom-chatgpt[.]com)
Conclusions
The observations highlight how high-profile trends and developments (such as the emergence and growth of AI) can have a transformative effect on the online infringement landscape and can drive an increased requirement for proactive brand protection programmes of monitoring and enforcement. This is particularly crucial for prominent brands in associated industry areas, whose names can readily be hijacked by infringers looking to take advantage of increased consumer interest and elevated levels of search queries, to drive traffic to their own content or make false claims of affiliation. Particularly where individual brands achieve a dominant position in the overall market, the brand owner may need to be mindful to proactively defend the space and enforce against infringements, to guard against the damaging effects of brand genericide and the resulting difficulties in subsequently achieving successful enforcement.
References
[1] 'Patterns in Brand Monitoring' by D.N. Barnett (Business Expert Press, 2025), Chapter 14: 'New developments'
[2] https://www.iamstobbs.com/opinion/the-top-generative-ai-brands-in-2024
[3] https://domainnamewire.com/2024/11/06/openai-acquires-chat-com/
[4] https://openai.com/index/chatgpt/, dated 30-Nov-2022
[5] 'Patterns in Brand Monitoring' by D.N. Barnett (Business Expert Press, 2025), Chapter 5: 'Prioritization criteria for specific types of content'
[6] https://circleid.com/posts/20230117-the-highest-threat-tlds-part-2
This article was first published on 10 December 2024 at:
https://www.iamstobbs.com/opinion/a-quick-chat.com-about-ai-domain-names
No comments:
Post a Comment