Ivy Estoesta Quoted in MIT Horizon: Generative AI, Copyright, and Infringement

Sterne Kessler’s Ivy Estoesta, director in the Mechanical & Design Practice Group and the Trademark & Brand Protection Practice, spoke to MIT Horizon on the uncertainty around generative artificial intelligence (AI) and intellectual property laws.

In the first article, “Are AI-Generated Works Copyrightable?” Estoesta explained that per the Copyright Office’s guidance, using AI as a tool would be eligible for copyright protections, while relying on AI as a generator would not. She further stated, “We don’t yet know the dividing line of when AI is used as a tool versus a generator.”

One issue AI developers face is proving their purpose is different enough from the scraped content licensed to other generative AI models. A judge could then determine that the AI developer’s purpose is the same as the creators, which is to train AI models. In a case featured in the second article article “Copyright Fight: Training Data and Infringement,” Getty Images sued Stability AI alleging the company used 12 million Getty images to train their image generator.

Estoesta explained that Getty Images, “indicated that they have licensed their images to another AI developer for training. If someone then scrapes the Getty website or database to do the same thing, the use is the same, and you’ve not changed the purpose.”

In another case, a federal judge ruled that AI startup, Ross Intelligence, infringed upon Thomson Reuters’s copyright. The AI company had taken legal summaries without permission from Thomson Reuters to train its AI driven legal research system. Estoesta commented on the outcome, “I think the case probably is going to affect how some AI companies are looking at their scraping practices,” and, “Maybe they are more likely to now obtain at least affirmative consent from the rights holders, and maybe they’ll get licenses.”

Estoesta spoke to the future for developers in the uncertain terrain of AI and copyright law, “I think AI developers would want a law, because it gives them more certainty about what they can do,” she continued, “And I imagine that someone who wants to go all in and invest in their AI software wants the assurance that they’ll be able to develop it and train it.”