A recent proposal advocating for the disclosure of copyrighted materials used in training conceptual AI systems has garnered support from numerous innovative business organizations.
California Democratic Representative Adam Schiff presented the Generative AI Copyright Disclosure Act to the House on April 9th.
Under this act, individuals involved in creating or modifying databases for AI training must inform the Register of Copyrights by providing a detailed overview of any copyrighted works utilized. This notification must be submitted at least 30 days before the AI technologies based on the dataset are made public. The legislation also covers conceptual AI systems developed in the past.
Rep. Schiff emphasized the transformative impact of AI on various aspects of society and stressed the importance of balancing its potential with regulatory measures. The Generative AI Copyright Disclosure Act aims to foster innovation while safeguarding the rights of creators, ensuring they are aware of their work’s contribution to AI training datasets. It underscores the significance of respecting creativity and harmonizing technological progress with ethical considerations.
The bill enjoys support from multiple trade associations in intellectual property-focused sectors, including leading music industry entities such as the Recording Industry Association of America (RIAA), the American Society of Composers, Authors, and Publishers (ASCAP), and various other prominent organizations.
Furthermore, endorsements come from the Human Artistry Campaign and entertainment industry groups like SAG-AFTRA, the Directors Guild of America, and the Writers Guild of America. These stakeholders emphasize the importance of transparent record-keeping in upholding creators’ rights and commend Rep. Schiff for championing this critical issue.
The proposed legislation aligns with similar initiatives in the European Union, such as the AI Act, which mandates transparency in training AI models with copyrighted materials. While the Generative AI Copyright Disclosure Act does not mandate licensing of copyrighted content, it addresses the need for accountability in AI development.
AI developers have cited fair use exemptions in copyright law to justify the use of copyrighted materials in training datasets. However, organizations like Fairly Trained are advocating for ethical practices in AI model development to uphold creators’ rights. Fairly Trained recently certified a large language model, signaling a positive shift towards responsible data usage in AI technology.
