true77
The U.S. Government’s Stance on AI-Generated Art and Copyright
Amidst a rapidly evolving AI market and growing concerns regarding the impact of AI on art creation, the U.S. government has established a significant standard: If a human element is absent, legal protections are not applicable. This stance was outlined in a comprehensive report on AI and copyright issued by the U.S. Copyright Office.
The report clarifies that outputs generated by AI tools without human intervention do not meet the criteria for federal copyright protection. The decision was reached as part of a long-term AI initiative aimed at addressing various legal uncertainties arising from the surge in AI development, such as the eligibility of AI-generated content for copyright protection under the U.S. Constitution’s Copyright Clause.
Human Involvement and Copyright Protection for AI-Generated Material
According to the report, generative AI outputs can only be eligible for copyright protection when sufficient expressive elements have been determined by a human author. This includes scenarios where a human-created work influences an AI output or where a human makes creative adjustments to the output, rather than solely providing prompts.
The report emphasizes the importance of maintaining the centrality of human creativity in works that incorporate generative AI in their creative process. Register of copyrights and U.S. Copyright Office director, Shira Perlmutter, stated that protecting materials predominantly determined by a machine would contradict the constitutional objectives of copyright.
For instance, images or videos generated by tools like Midjourney or OpenAI’s DALL-E 3 cannot be copyrighted by their creators, even if the individual provided a detailed and unique prompt for content generation. Sole prompts and subsequent iterations are also not eligible for copyright protection.
Guidelines for Human Involvement in AI Art Creation
The U.S. Copyright Office’s report also provides guidelines on the degree of human participation required in the creation of AI art utilizing assistive technologies, such as computer-generated images in filmmaking. A critical factor in their decision was the unpredictability of generative AI outputs, which can yield diverse results with similar or identical prompts.
A forthcoming third part of the report will delve into the role of copyright in training AI models and generative AI tools using original and protected materials.
