Companies from the entire ad ecosystem – from publishers and SSPs to advertisers and DSPs – utilize artificial intelligence (AI) in various ways to create content, improve operational efficiencies, and optimize campaigns. However, over the past two years, as multiple cases have been brought regarding the development, training, and usage of AI tools, particularly generative AI, companies must grapple with the significant and complex legal issues around such utilization. The U.S. Copyright Office and Federal courts have also begun to comment on what is, and is not, copyrightable. The Copyright Office also plans to release commentary on the training of AI models on copyrighted materials, and whether such training should be considered to fall into the ‘fair use’ exception to infringement claims. Digital advertising market participants should be aware of the intellectual property issues in using generative AI and what to watch for as litigation evolves on such matters. In particular, this panel will explore some of the technical, operational, and legal issues concerning the usage of generative AI, including:
- Copyrightability of prompts, outputs, and refined outputs
- Human authorship requirements in copyrights
- Data provenance (i.e., ability to trace the origin and history of the data)
- Control (i.e., implementing mechanisms to manage and regulate how AI systems are used)
- Fair use in light of the Supreme Court’s ruling in Warhol v. Goldsmith Supreme with a focus on training models and outputs generated from the models