VIRGINIA, UNITED STATES — Artificial intelligence (AI) research organization OpenAI has hit a roadblock after the United States Patent and Trademark Office (USPTO) issued a final refusal of their application to register “GPT” as a trademark.
GPT stands for “Generative Pre-trained Transformer,” representing OpenAI’s family of large language models (LLM).
The USPTO determined that “GPT” merely describes a feature or function of OpenAI’s AI software services. The trademark attorney reviewing OpenAI’s application stated that consumers would immediately recognize GPT as referring to key aspects of OpenAI’s goods and services related to generative AI and natural language processing.
“Accordingly, consumer encountering the acronym “GPT” would immediately understand this to communicate a feature of the applicant’s software good and services which are “generative pretrained transformers” or feature neural network models that “give applications the ability to create human-like text and content (images, music, and more), and answer questions in a conversational manner,” the refusal stated.
OpenAI originally filed the “GPT” application in March 2021, facing initial rejection in May 2022 on procedural grounds and a substantive denial in September 2022.
While termed a “final refusal,” OpenAI still has the option to request reconsideration within three months. The San Francisco-based AI lab could also pay a fee to extend the response deadline by an additional three months.
Although the trademark prospects remain unclear, OpenAI retains other intellectual property rights over its proprietary AI models and technologies. Securing a broad “GPT” trademark could still offer useful protections against potential confusion from unauthorized uses. The outcome has implications for OpenAI’s brand as well as the emerging AI industry.