OpenAl Dukes it Out with the USPTO over the CHATGPT Trademark

ChatGPT is one of the most popular AI chatbots in the world. However, the U.S. Patent and Trademark Office is unimpressed.

In 2022, ChatGPT’s supreme overlord owner, OpenAI, filed an application to register CHATGPT as a trademark for downloadable and online software.

The USPTO rejects applications for registration on the Principal Register if they are too descriptive. A trademark is too descriptive if it immediately conveys knowledge of a quality, feature, function, or characteristic of the goods/services with which it is used.

In the case of CHATGPT, the USPTO refused registration on the Principal Register because of the descriptive relationship between the mark and the software. According to the Examining Attorney assigned to the application, the term “CHAT” refers to a synchronous exchange of remarks over a computer network, and “GPT” is an acronym for “Generative Pre-trained Transformers,” a type of neural network model that generates human-like text. The Examining Attorney cites a variety of internet evidence showing that these terms are commonly used in OpenAI’s industry to describe similar types of software and services, leading to the conclusion that the mark lacks distinctiveness.

OpenAI recently appealed a final refusal while it tries to convince the Examining Attorney one last time that she is wrong about the distinctiveness of the trademark. And it has pulled out all the stops, submitting not one but two expert reports. Presumably, the point is to make all this part of the record for the appeal, as the Examining Attorney is unlikely to change her mind. It will then be up to the Trademark Trial and Appeal Board to decide this issue. Of course, given OpenAI’s unlimited resources, it’s unlikely to stop at the TTAB if things don’t go their way.

I asked ChatGPT what it thought the eventual outcome of this proceeding would be, but it played coy:

If OpenAI can effectively demonstrate that “ChatGPT” has become a source identifier and not just a description of a type of product, there is a reasonable chance of success. However, the strong descriptive refusal means that the path to registration is challenging, and the outcome is by no means guaranteed. If I had to make a prediction, I’d say the case could go either way, but a successful registration will require a well-prepared and robust case from OpenAI.

So there you have it: the case could go either way. Spoken like a true AI lawyer.