AI Giants Advocate for Fair Use in Copyright Law Amidst Generative AI Debate
Major artificial intelligence (AI) companies have come together to advocate for fair use in copyright law amidst the ongoing debate surrounding generative AI. The U.S. Copyright Office is currently seeking perspectives on how copyright and intellectual property should be handled in relation to generative AI, and prominent AI companies have shared their opinions on the matter.
One of the companies, Meta, provided a detailed letter in late October 2023, explaining the background of AI development and the use of training data. Meta argues that the use of copyrighted material to train generative AI models does not violate copyright law as it is not a consumptive use. Even if copyright protection is triggered, Meta believes it falls under fair use. According to Meta, training AI models is a transformative process that extracts statistical information from language and abstract concepts to generate new content. The company clarifies that AI models do not store copyrighted data but rather learn patterns and relationships from the training data, thereby not violating the rights of copyright holders.
Meta also highlights several issues with proposed legal licensing mechanisms and asserts that copyright law should not protect artistic style. While there are concerns about AI systems imitating artists’ voices, looks, or styles, Meta argues that existing state publicity laws, federal unfair competition laws, and First Amendment principles adequately cover these concerns. Hence, Meta suggests that no significant changes to current law are necessary to regulate AI.
Furthermore, Meta views generative AI as a tool that enhances human creativity and productivity, comparing it to a printing press, a camera, or a computer. The company warns that licensing AI training data on the necessary scale would be exorbitantly expensive and could halt the progress of generative AI development. Negotiating licensing agreements with individual rights holders would only cover a small fraction of the required data, making it administratively impossible to locate all rights holders for content such as online reviews. Similar arguments have been put forth by OpenAI and Google, highlighting the flexibility of existing copyright principles in dealing with AI scenarios.
Google, like Meta, asserts that content generated by AI without human intervention is not copyrightable. However, if human intervention and creativity are involved in the generative AI process, copyright protection may be granted. Google argues that a work is considered infringing only if it is substantially similar to the allegedly copied work. While this possibility cannot be ruled out for generative AI, Google believes it is unlikely. The company emphasizes the need to strike a balance between the interests of rights holders and the public, cautioning against premature legislative action that may stifle innovation and limit the potential of AI technology.
OpenAI also supports fair use and argues that generative AI does not reproduce copyrighted material extensively. It compares the learning process of AI models to that of a human child, abstracting factual metadata to create new and original content. OpenAI, like Google, refers to blocking data crawlers to prevent unintended inclusion of specific data in future AI models. However, this option only affects future models and has no impact on existing models or datasets.
Apple, with its focus on generative AI tools for program code, states that automating the development of computer programs is not a new development and that AI coding tools represent a significant evolution of this process. The company argues that when a human developer controls the generative AI tools and determines the form and usage of the produced code, it has sufficient human authorship to be protected by copyright.
In conclusion, major AI companies advocate for fair use in copyright law regarding generative AI. They argue that generative AI is a transformative use of data and does not extensively reproduce copyrighted material. The companies caution against premature legislative action, emphasizing the need for a balanced approach that supports innovation while respecting the interests of rights holders and the public. The ongoing debate will shape the future of copyright law in the context of AI technology.
References:
– [THE DECODER](https://www.thedecoder.in/)
– [Meta](https://www.meta.com)
– [Google](https://www.google.com)
– [OpenAI](https://openai.com)
– [Apple](https://www.apple.com)
