Appellants Say Court’s OpenAI Secondary Meaning Ruling Was In Error

·

(May 16, 2024, 10:34 AM EDT) -- SAN FRANCISCO — A trial court erred in finding that the OpenAI mark acquired a secondary meaning with the release of its Dall-E website and before the release of the vastly more popular ChatGPT while ignoring analogous uses of the mark and that all the allegedly irreparable harm was speculative, a company tells the Ninth Circuit U.S. Court of Appeals.

(Open Artificial Intelligence, et al. v. OpenAI, Inc., No. 24-1963, 9th Cir.)

(Opening brief available.  Document #46-240605-016B.)

The appellants filed their brief on May 8.

In a complaint filed in August in the U.S. District Court for the Northern District of California, OpenAI says that soon after it announced its founding, Guy Ravine sought to trademark the term “Open AI” with the U.S. Patent and Trademark Office (USPTO).  OpenAI says Ravine is now actively interfering with its own attempts to register its mark with the USPTO.  OpenAI alleges trademark infringement and unfair competition under Title 15 U.S. Code Section 1125(a), 15 U.S.C. § 1125(a), common-law trademark infringement and fraudulent registration under Title 15 U.S. Code Section 1120, 15 U.S.C. § 1120, and cancellation–no bona fide use and cancellation–misrepresenting source under Title 15 U.S. Code Section 1119, 15 U.S.C. § 1119.

Judge Yvonne Gonzalez Rogers granted OpenAI’s motion for an injunction on Feb. 28, finding that “the landscape has changed” in the wake of OpenAI launching two products that have made it a household name.  Judge Rogers rejected Ravine’s argument that he was the senior user of the mark.

In a March 27 motion for reconsideration filed by Ravine and his company, Open Artificial Intelligence Inc., they argued that the judge erred in finding the OpenAI mark descriptive.

Secondary Meaning

While the judge used the term “suggestive,” it appears that she was indicating that the mark acquired a secondary meaning, the defendants said.  But the USPTO found that the mark had not acquired a secondary meaning in both January 2023 and April 2023.  While there is evidence that certain terms had acquired a secondary meaning by September 2022, there is no evidence that OpenAI was one of them, the defendants argued.  “This Court’s finding that Plaintiff’s mark likely acquired secondary meaning by September 2022 is based on the clearly erroneous factual premise that the USPTO last determined the mark lacked secondary meaning months earlier,” the defendants argued.  And nowhere in the briefing on injunctive relief do the plaintiffs argue that the mark had acquired a secondary meaning, the defendants said.

In her April 5 ruling denying the motion, Judge Rogers said that she gave the defendants every opportunity to respond to the unrefuted evidence but that they did not avail themselves of that chance.  That evidence shows that Ravine and Open Artificial Intelligence were never actual users of the OpenAI trademark, Judge Rogers said.

Open Artificial Intelligence and Ravine appealed to the Ninth Circuit.

Timing

In their opening brief, the appellants argue that the injunction should be reversed because OpenAI has not established that it is likely to succeed in showing that its OpenAI mark acquired a secondary meaning in September 2022.  Assuming that OpenAI acquired a secondary meaning in September 2022 requires concluding that it acquired that meaning based on the release of the less popular Dall-E website and before the release of ChatGPT.  That conclusion “is simply not supported by the record or reason,” the appellants say.

Secondly, the court erroneously granted a permanent injunction, the appellants say.  A permanent injunction requires a showing that the facts clearly favor the recipient.  But the court below only relied on the less rigorous likelihood-of-success standard, the appellants say.

The appellants say the trial court erred in not considering the analogous use doctrine.  The appellants say there is no question that they registered the open.ai domain name and that even accepting the evidence as the trial court did, there is undisputed evidence of the mark’s analogous use.

As to harm, the District Court improperly applied a presumption of irreparable harm given that success on the merits is unlikely, the appellants say.  But in any case, finding irreparable harm based on speculative statements was improper.  The party seeking the injunction may not simply speculate that an injury is possible, it must demonstrate the likelihood of the threatened injury, the appellants say.

“Finally, the preliminary injunction is overbroad to the extent that it prohibits non-infringing uses of Ravine Parties’ open.ai domain.  That is not necessary to protect against consumer confusion and it improperly prohibits legal uses of the open.ai domain that do not use the disputed mark,” the appellants tell the court.

Counsel

OpenAI is represented by Robert M. Schwartz and Aaron H. Perahia of Quinn, Emanuel, Urquhart & Sullivan LLP in Los Angeles, Sam S. Stake of the firm’s San Francisco office and Margret M. Caruso of its Redwood Shores, Calif., office.

The defendants are represented Ryan G. Baker, Scott M. Malzahn and Patricia Rojas-Castro of Waymaker LLP in Los Angeles.

(Additional documents available:  Order on reconsideration.  Document #46-240501-015R.  Order.  Document #46-240403-003R.)