TRAIN Act Targets Transparency in Generative AI Training Practices
Representatives Madeleine Dean (D-PA) and Nathaniel Moran (R-TX) have introduced a bipartisan bill known as the Transparency and Responsibility for Artificial Intelligence Networks (TRAIN) Act in the US House of Representatives. This legislative initiative aims to establish a framework to help musicians, artists, writers, and other creators ascertain if their copyrighted work was used to train generative artificial intelligence (AI) models without their consent and, if so, to allow them to pursue compensation.
Legal Mechanism for Copyright Owners
The proposed legislation introduces a new legal mechanism that empowers copyright owners to utilize federal court subpoena power to obtain information regarding the materials used to train generative AI models. This bill seeks to establish the first federal statutory definition of “generative AI models”. A generative AI model is defined as an AI model that “emulates the structure and characteristics of input data to generate derived synthetic content,” such as images, video, audio, text, and other forms of digital content.
Subpoena Process
Under the proposed framework, copyright owners would be eligible to request a subpoena compelling a generative AI developer to provide copies or records sufficient to identify the copyrighted works used in training the model. To obtain a subpoena, the copyright owner must submit a request to a US district court clerk that includes a proposed subpoena and a sworn affidavit affirming their good faith belief that their copyrighted works were used to train the developer’s model.
The affidavit must also confirm that the purpose of the subpoena is to obtain the copyrighted materials or records needed to protect the copyright owner’s rights. Once issued, the developer is mandated to “expeditiously disclose” the requested copies or records to the copyright owner or an authorized representative.
Enforcement Provisions
The TRAIN Act incorporates two key enforcement provisions. First, if a developer fails to comply with a subpoena, the court may apply a rebuttable presumption that the developer did use the copyrighted works to train its model. Second, if a copyright owner seeks a subpoena in bad faith, the recipient has the right to pursue sanctions under Federal Rule of Civil Procedure 11.
Impact of the TRAIN Act
If enacted, the TRAIN Act would significantly enhance transparency obligations for developers of generative AI models while providing copyright owners a valuable tool to investigate potential unauthorized uses of their works. This legislative move underscores the growing need for accountability and clarity in the rapidly evolving field of generative AI.
