(CTN News) – It’s understandable that OpenAI feels protective of its brand. Recent trademark applications filed with the USPTO include ThreatGPT, MedicalGPT, DateGPT, and DirtyGPT.
All are based on ChatGPT, the chatbot OpenAI launched in November using its deep learning model, GPT-4, whose latest version was released last month.
After applying for a trademark in December for “GPT,” which stands for “Generative Pre-trained Transformer,” OpenAI petitioned the USPTO last month to expedite the process, citing the “myriad infringements and counterfeit apps” starting to emerge.
OpenAI’s petition last week got dismissed.
They also didn’t provide “appropriate documentary evidence to support special action,” the agency said.
OpenAI may have to wait up to five more months for a decision, based on the rest of the queue, says Jefferson Scher, a partner in Carr & Ferrell’s trademark practice group. In any case, Scher says, the outcome isn’t certain.
It is certainly possible for OpenAI to secure the patent, he says. He was asked, for example, whether OpenAI might face resistance because the “T” in GPT stands for “Transformer,” the name of Google’s neural network architecture unveiled in 2017 and now widely used.
If GPT has such a descriptive origin, can it still be a brand? Scher wonders. He points to IBM, short for International Business Machines, as just one example of a brand that has a descriptive origin, even if it is weak.
Scher emphasizes that such precedents are not a guarantee that will own GPT in the future, but they do provide some support.
Moreover, Scher points out that has been using “GPT” for years, having released its original Generative Pretrained Transformer model, or GPT-1, in October 2018.
However, Scher noted that it is a “funny situation,” since “normally, when you base claims on use, you’ve built up your brand in the marketplace over time.” OpenAI was mostly known to AI researchers until last year, when the company became a kind of overnight sensation when it released a mesmerizing deep learning model (DALL-E 2), followed by ChatGPT.
The “GPT” trademark application will go through a so-called opposition process afterward, in which other market participants can argue why the USPTO should deny it.
According to Scher, OpenAI must establish that “GPT” is proprietary rather than perceiving it as a generic term for generative AI in the public’s mind.
What would be the USPTO’s ruling on public perception? “One scenario would be to take a random sample of Americans and ask them to answer the question,” says Scher, but the government refuses to fund such a project, so anyone challenging OpenAI would have to pay.
The use of “GPT” in public, such as late-night talk shows and public writing, is another way to establish public perception. A trademark trial would determine if it is protectable if people are not treating it as proprietary.
This would entail a lengthy process, which OpenAI certainly doesn’t want.
The company might be able to take advantage of it in this long process. The more time passes, the more users OpenAI accrues, and the more coverage the company receives, the more likely that last scenario becomes.
What is the average household’s knowledge of OpenAI? The government may be closing in on that, says Scher.