As companies use large data reservoirs to train generative AI platforms, concerns around the infringement of Intellectual Property (IP) are growing. With recent initiatives to scrutinise AI developers in the US, policymakers are emphasising the need to strike a balance between nurturing the AI industry and safeguarding IP (Kline, 2024).
Copyright laws have not kept pace with the rapidly evolving AI platforms (Forbes Breaking News, 2024). However, challenging questions on the development of generative AI make it tricky to craft measures that safeguard IP, while fostering an emerging yet evolving industry. For instance, questions on the ownership of AI-produced content make it hard to pin responsibility for copyright violations. There is also a debate on whether the training of AI should be allowed when copyrighted and trademarked work is directly referenced albeit without permission (Appel, Neelbauer, & Schweidel, 2023).
As these deliberations unfold, a series of legal cases (~24 lawsuits) against generative AI platforms surfaced, indicating a growing demand for the scrutiny of AI companies’ training methods. For instance, in 2022, three artists brought a suit against numerous platforms for using their original work and styles to train generative AI. They claimed that the generative AIs’ works were unauthorised derivatives since these were inadequately transformative relative to the original works (Chen, 2023). In 2023, The New York Times, sued Microsoft and OpenAI for violating copyrights, claiming that the news outlet’s articles were used unlawfully for training purposes (Grynbaum & Mac, 2023).
‘Crystalline Maples’ is part of Erin Hanson’s body of work that was used to train platforms like Stable Diffusion, without consent or compensation. The outputs generated by the platform strikingly resemble her original work.
The Generative AI Copyright Act : Provisions
To address rising claims of rampant copyright infringement by these platforms, Adam Schiff, the US representative for California, introduced a Bill titled ‘Generative AI Copyright Disclosure Act’ in April 2024 (Forbes Breaking News, 2024). Multiple professional organisations including the Recording Industry Association of America and Writers Guild of America supported this initiative.
If the bill is enacted, companies yet to launch their platform shall be mandated to provide details of copyrighted content used for training purposes 30 days before their public launch. On the other hand, companies that have launched their platforms already shall be required to provide these details 30 days after the law is effective. If companies fail to submit the necessary notice within the stated time frame, they will have to bear a penalty of at least $5000. Further, the Copyright Office would be expected to maintain an exhaustive database with publicly available information on the notices submitted (Kline, 2024).
Implications of enaction
Consent to use and compensation for original work is especially important for various creators including artists and writers. The extent to which copyright holders’ intellectual property is used can be assessed through a public database, potentially influencing their decision to revise licenses or seek compensation from those training generative AI (Holloway et al., 2024; Kline, 2024).
AI developers are likely to witness challenges should this bill be passed. Smaller AI companies or startups especially may be deterred from entering this market. They are likely to be under-equipped to bear legal liabilities for failing to submit comprehensive records of all copyrighted work within the stipulated time frame, or bear the technical and administrative costs tied to maintaining these records.
On the other hand, larger companies like OpenAI may continue to defend their training methods by citing “fair use”. Representatives of OpenAI previously suggested that utilising publicly available material is necessary for innovation and competitiveness; denying that they were being unfair to creators. The “fair use” provision, detailed under Section 107 of the Copyright Act, allows the reproduction of work for news reporting, teaching, and critiquing, for instance. Broadly, it is meant to support the flow of information critical to ‘public interest’ (Yoon & Borges, 2024; Lobo, 2024).
The legal proceedings against leading AI companies and emerging legislation in the US allow other countries insights into the challenges and pathways to regulate an evolving industry.
Resources:
Appel, G., Neelbauer, J., & Schweidel, D. A. (2023, April 7). Generative AI Has an Intellectual Property Problem. Retrieved from Harvard Business Review website: https://hbr.org/2023/04/generative-ai-has-an-intellectual-property-problem
Forbes Breaking News. (2024, April 13). Adam Schiff Discusses AI Threat To IP Protections, Promotes “Generative AI Copyright Disclosure Act.” Retrieved June 3, 2024, from YouTube website: https://www.youtube.com/watch?v=dhfG8iyYCnE
Harvard University. (2016). Copyright and Fair Use. Harvard.edu. https://ogc.harvard.edu/pages/copyright-and-fair-use
Holloway, J., Cheng, M., & Dickenson, J. S. (2024, January 13). Using generative AI and protecting against copyright issues. World Economic Forum. https://www.weforum.org/agenda/2024/01/cracking-the-code-generative-ai-and-intellectual-property/
Kline, D. (2024). Understanding Proposed Generative AI Copyright Disclosure Act of. Retrieved June 3, 2024, from natlawreview.com website: https://natlawreview.com/article/generative-ai-copyright-disclosure-act-2024-balancing-innovation-and-ip-rights
Grynbaum, M. M., & Mac, R. (2023, December 27). The Times Sues OpenAI and Microsoft Over A.I. Use of Copyrighted Work. The New York Times. Retrieved from https://www.nytimes.com/2023/12/27/business/media/new-york-times-open-ai-microsoft-lawsuit.html
Lemley, M. A. (2023, July 21). How Generative AI Turns Copyright Law on its Head. https://doi.org/10.2139/ssrn.4517702
Lobo, S. (2024, April 10). New US bill will require AI companies disclose copyrighted material. MEDIANAMA.https://www.medianama.com/2024/04/223-us-proposes-new-bill-that-will-require-ai-companies-to-disclose-use-of-copyrighted-material/
Stanford Law School. (2024, April 22). FutureLaw 2024 – Generative AI and Intellectual Property. Retrieved June 3, 2024, from YouTube website: https://www.youtube.com/watch?v=AT3IEgsC1dA
Yoon, J., & Borges, C. (2024). Informing the Innovation Policy Debate: Key Concepts in Copyright Laws for Generative AI | Perspectives on Innovation | CSIS. Www.csis.org. https://www.csis.org/blogs/perspectives-innovation/informing-innovation-policy-debate-key-concepts-copyright-laws
Banner image source: Adobe Firefly
