Go offline with the Player FM app!
Season 7, Episode 10: LLMs as Uncopyrightable Idea Extraction and Generation (with Damien Riehl)
Manage episode 431379599 series 2967955
In this episode of Notes to my (Legal) Self, dive into the captivating world of Large Language Models (LLMs) with legal expert Damien Riehl. Explore how LLMs are transforming the landscape of idea extraction and generation, and the intricate legal challenges they pose.
Expect an engaging discussion on the significant lawsuits against LLM companies, such as NY Times v. Microsoft/OpenAI and Doe v. Github, and why major players like Google, Microsoft, and Adobe are indemnifying LLM users against copyright lawsuits. Gain insights into the legal foundations of the idea-expression distinction, understanding how neural nets process text into uncopyrightable vector embeddings, and why LLM inputs and outputs often present strong fair use defenses.
Damien provides an in-depth analysis of how the idea-expression dichotomy is at the heart of many copyright cases, arguing that ideas themselves are not copyrightable—only their expressions are. He explains that LLMs extract ideas from vast data sets and generate new expressions, thus navigating away from copyright infringement claims. The conversation delves into the complexities of determining what constitutes an "original" work and the implications of statistical likelihood in LLM-generated content.
In the practical realm, discover why Big Tech's support for LLM users is a strategic move to avoid bad case law, and learn how in-house lawyers can protect their companies when relying on these technologies. Explore the future of business models in legal tech, where Damien predicts that the value will lie in unique data ("oil") and innovative products, rather than simply wrapping LLMs in new interfaces.
Key takeaways from this episode include:
- Grasp the legal reasoning behind Big Tech’s support for LLM users.
- Master the idea-expression dichotomy for copyright eligibility.
- Learn about the transformation of text into uncopyrightable ideas via LLMs.
- Discover how LLM inputs and outputs navigate away from infringement claims.
Whether you're an attorney, a legal technologist, or simply intrigued by the intersection of law and technology, this episode offers valuable insights into the evolving landscape of copyright risks and AI. Don’t miss this chance to stay ahead in the rapidly changing world of legal innovation.
140 episodes
Manage episode 431379599 series 2967955
In this episode of Notes to my (Legal) Self, dive into the captivating world of Large Language Models (LLMs) with legal expert Damien Riehl. Explore how LLMs are transforming the landscape of idea extraction and generation, and the intricate legal challenges they pose.
Expect an engaging discussion on the significant lawsuits against LLM companies, such as NY Times v. Microsoft/OpenAI and Doe v. Github, and why major players like Google, Microsoft, and Adobe are indemnifying LLM users against copyright lawsuits. Gain insights into the legal foundations of the idea-expression distinction, understanding how neural nets process text into uncopyrightable vector embeddings, and why LLM inputs and outputs often present strong fair use defenses.
Damien provides an in-depth analysis of how the idea-expression dichotomy is at the heart of many copyright cases, arguing that ideas themselves are not copyrightable—only their expressions are. He explains that LLMs extract ideas from vast data sets and generate new expressions, thus navigating away from copyright infringement claims. The conversation delves into the complexities of determining what constitutes an "original" work and the implications of statistical likelihood in LLM-generated content.
In the practical realm, discover why Big Tech's support for LLM users is a strategic move to avoid bad case law, and learn how in-house lawyers can protect their companies when relying on these technologies. Explore the future of business models in legal tech, where Damien predicts that the value will lie in unique data ("oil") and innovative products, rather than simply wrapping LLMs in new interfaces.
Key takeaways from this episode include:
- Grasp the legal reasoning behind Big Tech’s support for LLM users.
- Master the idea-expression dichotomy for copyright eligibility.
- Learn about the transformation of text into uncopyrightable ideas via LLMs.
- Discover how LLM inputs and outputs navigate away from infringement claims.
Whether you're an attorney, a legal technologist, or simply intrigued by the intersection of law and technology, this episode offers valuable insights into the evolving landscape of copyright risks and AI. Don’t miss this chance to stay ahead in the rapidly changing world of legal innovation.
140 episodes
All episodes
×Welcome to Player FM!
Player FM is scanning the web for high-quality podcasts for you to enjoy right now. It's the best podcast app and works on Android, iPhone, and the web. Signup to sync subscriptions across devices.