In the Bloomberg Law article “European AI Act Training Disclosures Expose US Copyright Risks,” partner Mauricio Uribe discussed how the European Union’s Artificial Intelligence Act, set to go into effect this week, could impact companies in the United States serving customers in the EU
In part, the article covers a key section of the Act which requires providers of general-purpose Large Language Models (LLMs), such as Chat GPT-4, to publicly share details on the content used to train the AI model. Speaking to the point that this provision could lead to litigation from copyright holders who believe AI model training is an unlawful use of their intellectual property, Uribe maintained that this would likely only affect cases in which litigation had not already begun. In the case of ongoing litigation, he explained, the use of copyrighted content for training would have already been made clear in the discovery process.
The article examined the impact the AI Act could have on companies who must undertake the laborious task of uncovering all the data used to train their models in order to be compliant with the Act’s provisions by the applicable deadline. While the earliest deadline is August 2025, Uribe emphasized that companies should begin their compliance work sooner rather than later. “Even though you don’t have to do something now,” he said, “start early.”
Read the full article here.
Earlier this year, Uribe led a discussion on the Knobbe IP+ podcast with European patent attorney and Ph.D. Hanane Fathi Roswell regarding steps companies should take to prepare for the EU AI Act. You may listen to their discussion here.
