Generative AI and Branding: Why Businesses Cannot Afford to Skip Trademark Clearance

Legal Alert

August 17, 2026

Generative AI and Branding: Why Businesses Cannot Afford to Skip Trademark Clearance

Without a doubt, generative AI is rapidly becoming a staple in the branding industry. The appeal is clear: lower costs, faster turnaround times, and seemingly endless creative possibilities.

Although generative AI is transforming this process, several core trademark laws have not changed, and can lead to infringement claims, forced rebranding, and lost trademark rights.

Getty Images v. Stability AI: The UK Decision and Parallel U.S. Litigation

In January 2023, stock photography company Getty Images sued Stability AI, the developer of the Stable Diffusion image-generation model, in the High Court of England and Wales. The debate centered on whether training the AI on copyrighted images was lawful, but the trademark claims proved equally formidable.

Getty alleged that Stability AI’s models frequently produced synthetic outputs featuring distorted or garbled versions of the iconic Getty Images and iStock watermarks. Getty claimed this constituted trademark infringement, false designation of origin, and trademark dilution. Getty argued that the presence of these distorted watermarks falsely implied that Getty had endorsed or sponsored the AI-generated images or was otherwise associated with them.

Stability AI attempted to dismiss the trademark claims, arguing that the AI-generated images did not constitute a "use in commerce" and framing the dispute as a copyright claim barred by previous Supreme Court precedent.

So, what did the High Court decide? On November 4, 2025, the High Court of England and Wales rejected Getty’s remaining secondary copyright claim, holding that Stable Diffusion’s model weights were not ‘infringing copies’ because the model did not contain or reproduce copies of the asserted copyrighted works. The court nevertheless found limited trademark infringement involving certain Getty Images and iStock watermark-like outputs. Importantly, the ruling did not decide whether training an AI model on copyrighted works in the UK would itself constitute infringement because Getty had abandoned its primary copyright and database-right claims during trial.

Meanwhile, Getty is pursuing a parallel action against Stability AI in the U.S. District Court for the Northern District of California. That case, Getty Images (US), Inc. v. Stability AI Ltd., No. 3:25-cv-06891-TLT-SK, remains at an early stage, with Stability AI filing its amended answer in July 2026. Although the UK decision is not binding on U.S. courts, it may provide an early indication of how courts will approach similar copyright and trademark questions involving generative AI in the United States.

The aftermath of the ruling sparked calls from UK content creators and legal experts for governance and transparency rules concerning AI data scraping. Interestingly, in June 2026, Getty Images announced its partnership with OpenAI, licensing content directly for AI search tools following the court ruling.

The Aftermath & What’s Next for Generative AI in Trademark Litigation

The evolution of cases like Getty Images v. Stability AI directly impacts how professionals and businesses should handle digital assets and AI tools. Accordingly, the absence of an obvious watermark on an AI-generated image or brand asset does not mean that the asset is free of third-party rights or potentially infringing trade dress similarities. Furthermore, pulling random images directly from the internet or relying on AI tools may open a business to hidden trademark, trade dress, or false attribution claims, especially if proprietary watermarks or recognizable brand elements leak into the generation process.

Legally, organizations are advised to treat AI outputs as a creative starting point. From a practical standpoint, businesses should review post-generation screening filters and maintain auditable documentation concerning how their assets were generated. This includes conducting a comprehensive, professional trademark clearance search before adopting any AI-generated name, logo, or slogan to ensure that the business does not inadvertently infringe existing common-law or registered marks.

Finally, companies should review the terms of service of the specific AI tools they use to confirm they hold the appropriate commercial rights, and draft clear agreements with marketing agencies or third-party vendors that explicitly outline liability, ownership, and indemnification for AI-generated content.

The biggest trademark risks in AI content are the ones you do not notice at first glance. A stray watermark or a recognizable piece of trade dress can turn a quick, low-cost asset into an expensive rebrand. When in doubt, clear it before you use it.

Interested in reading about the next landmark case?

Keep an eye out for Encyclopedia Britannica, Inc. and Merriam-Webster, Inc. v. OpenAI, filed in March earlier this year.

The famous publishers invoked the Lanham Act, stating that ChatGPT frequently generates "hallucinations" (fabricated or false information) and incorrectly attributes those false answers to Britannica and Merriam-Webster. The publishers contend that displaying their allegedly famous marks alongside hallucinated or otherwise incorrect text constitutes false designation of origin and trademark dilution. Furthermore, the publishers allege that OpenAI unlawfully scraped nearly 100,000 copyrighted reference articles and dictionary entries to train its AI models.

The case (No. 1:26-cv-02097) is still in its early stages with fact discovery and pretrial proceedings underway before Judge Sidney H. Stein.