The opposition, filed in October 2023 under Sections 5(2)(b) and 5(3) of the Trade Marks Act 1994, relied on a suite of earlier marks including CHATGPT, GPT, GPT-3, GPT-4 and GPT-5.

Comparison of the marks

The hearing officer came to the conclusion that CHATGPT had the best chance of a finding of similarity with CallGPT; consequently, despite all of the relied-upon marks being compared, the judgment largely focused on the CHATGPT mark.

Visually, the CallGPT and CHATGPT marks were held to share a medium to high degree of similarity, driven by the shared initial “C” and the identical “GPT” suffix. Aurally, similarity was assessed as medium due to the identical pronunciation of GPT.

Conceptually, the marks were found to have a medium degree of similarity, as “CALL” and “CHAT” both evoke forms of communication. Notably, however, the hearing officer concluded that “GPT” would not convey any clear meaning to the average UK consumer, despite the applicant’s assertions that GPT is a widely used descriptor for “generative pre-trained transformers”.

Distinctiveness

CHATGPT was found to possess a low to medium degree of inherent distinctiveness, reflecting the allusive nature of “CHAT” in the context of natural language processing services.

Yet more significantly, the hearing officer maintained their position that “GPT” lacked a clear meaning for the average consumer. This conclusion was reached despite the applicant providing evidence in the form of weblinks and printouts of webpages which refer to “generative pre-trained transformers”, including a webpage that specifically explains what the acronym stands for.

However, these materials were not considered sufficient by the hearing officer, criticising the reliance on sources such as Wikipedia, which is “inherently unreliable”, and the lack of evidence indicating how widely known this meaning would be amongst average UK customers.

Despite the hearing officer’s conclusion, it is unlikely that this establishes “GPT” as inherently non-descriptive in all contexts. Rather, it may underscore the evidential shortcomings in the material provided by the applicant, serving as a reminder of the need to provide robust evidence from various sources, even when a point may feel obvious from an insider perspective.

On the evidence, the hearing officer also found that CHATGPT had acquired enhanced distinctiveness, to a medium to high degree, in relation to specific class 9 and 42 goods/services concerning natural language processing. This conclusion was exclusively driven by substantial evidence of direct UK user engagement, including millions of website visits and app downloads, which highlights the persuasive value of clear quantitative, market-facing data in establishing acquired distinctiveness.

Comparison of goods and services

The applied for goods and services, all relating to software, were found to range from a low to high degree of similarity with those covered by CHATGPT. Notably, goods/services connected to artificial intelligence were considered to share a higher degree of similarity.

Likelihood of confusion

Taking into account the similarity of the marks, the overlap in goods and services, and the enhanced distinctiveness of CHATGPT, the hearing officer found a likelihood of both direct and indirect confusion.

A key observation was that “the average consumer tends to see what they expect to see.” In this context, the strong recognition of CHATGPT increased the likelihood that similar signs used for similar services would be mistaken for it.

Importantly, the hearing officer also identified a likelihood of indirect confusion. The substitution of “CHAT” with “CALL” was considered a logical variation that was consistent with brand extension. Consumers could reasonably assume that CALLGPT represents a variant or sub-brand of CHATGPT.

On the face of it, this may suggest that “GPT” operates as a core brand identifier of OpenAI, yet the outcome may equally reflect the applicant’s failure to convincingly provide evidence that “GPT” is a widely understood descriptive term, highlighting the ripple effect poor evidence can have.

In light of these findings, the opposition under s5(2)(b) succeeded.

Reputation

Having already established similarity between the marks, the hearing officer turned the s5(3) ground and whether CHATGPT enjoyed a reputation among the relevant public. Applying the same factors used to assess enhanced distinctiveness, they readily concluded that it did.

This finding was again grounded in the strong evidence of UK user engagement, including substantial website traffic and app downloads. The reputation of the mark was further supported by coverage in major UK media outlets such as The Guardian and The Times, reinforcing its prominence beyond its immediate user base.

Link

The hearing officer had little difficulty in finding that the relevant public would establish a link between CALLGPT and CHATGPT. This was driven by the combination of the strong reputation of CHATGPT and the similarity between the signs, making it likely that customers would mentally associate the two marks.

Damage (unfair advantage)
The hearing officer concluded that use of CallGPT would confer an unfair advantage on the applicant, due to the benefit they would gain from CHATGPT’s strong reputation.

The hearing officer particularly highlighted articles, provided by the opponent, which showed the CHATGPT mark in the context of advancements of AI. Consequently, they considered the mark to have a reputation for being at the forefront of technology, which would transfer to the applicant, allowing them to unfairly benefit.

Accordingly, all elements required to establish a successful opposition under s5(3) were satisfied.

Conclusions

The central takeaway from this decision is that outcomes in trade mark oppositions are often determined as much by the quality of evidence and representation as by the marks themselves.

On the applicant’s side, a decisive weakness lay in the failure to substantiate the claim that “GPT” had a commonly understood meaning among average consumers. This evidential gap therefore prevented the applicant from demonstrating any reduction in the inherent distinctiveness of the opponent’s mark, conceptual dissimilarity and ultimately reinforced the hearing officer’s finding of indirect confusion. As a result, the lack of varied and reliable evidence destroyed any hope the applicant had of successfully defending the opposition.

Consequently, this case serves as a cautionary example for parties engaging in opposition proceedings without professional representation, as even potentially arguable points will likely fail if not properly evidenced.

By contrast, through professional representation and comprehensive materials, including national press coverage and user base figures, the opponent was able to demonstrate that CHATGPT had acquired an enhanced level of distinctiveness and enjoyed a strong reputation. This foundation enabled the hearing officer to accept the opposition on both grounds with relative ease. 

Case details at a glance

Jurisdiction: England & Wales
Decision level: UKIPO
Parties: XEROTECH LTD and Noman Ahmed Shah v OpenAI OPCO LLC
Date: 16 March 2026 
Citation: O/0219/26
Decision: dycip.com/ukipo-openai-xerotech