The “To Keep Up” Wiki

CRISPR, AI, and Patent Law

Dick Wagner first presented an introduction to CRISPR technology and its patent disputes between different research teams, including historical context and current patent landscapes in the US and Europe. The discussion then shifted to copyright law fundamentals, covering basic concepts, eligible works, and defenses against infringement, with specific examples of fair use cases and recent legal decisions. The final portion focused on copyright and AI interactions, including discussions about AI-generated artwork, the use of pirated content, and the implications of AI technology on content creators and publishers.

Dick gave an introduction with his background in physics, biotech, and law, and explained the basics of CRISPR technology. He then discussed the ongoing patent disputes between Jennifer Doudna and Emmanuel Charpentier's team, and Feng Zhang's team, noting that Zhang's patent is currently dominant in the US while Doudna and Charpentier's patents are dominant in Europe. DIck explained that these disputes have been ongoing for over a decade, with the outcome still uncertain as of May 2023.

The discussion focused on the historical context of CRISPR development, noting that Zhang and Charpentier corresponded but did not directly collaborate, while Isaacson's book details their interactions. Carl inquired about the overlap between two patents covering CRISPR's use in eukaryotic organisms, which Dick confirmed. The conversation continued with an introduction to copyright issues in AI, highlighting ongoing legal disputes between tech companies and content creators, including a recent settlement proposal by Anthropic regarding their use of pirated books for training their AI system.

Dick explained the legal basis and structure of copyright law in the United States, highlighting its constitutional foundation and the distinction between economic and moral rights. He discussed the exclusive rights granted to copyright holders, including reproduction, distribution, adaptation, display, and performance, and emphasized the importance of fixing creative works in a tangible form. He also covered eligible works for copyright, such as literary, musical, and artistic creations, as well as defenses against copyright infringement, including independent creation and fair use. He recommended filing copyrights with the U.S. Copyright Office for legal protection and outlined the duration of copyright protection, which extends for the life of the author plus 70 years.

The discussion covered copyright law, focusing on the first AI-generated image that obtained a copyright, which involved human modifications to make it acceptable. Dick explained that AI alone cannot be copyrighted, requiring human intervention to create copyrightable material. He also discussed derivative works, explaining that these are copyrightable but require permission from the original copyright holder, using examples like translations, adaptations, and compilations.

Dick explained the concept of derivative works in copyright law, emphasizing that transformations must be substantial to be copyrightable. He discussed the four factors considered in determining fair use: (1) the purpose of use, (2) the nature of the copyrighted work, (3) the amount and substantiality of the portion used, and (4) the effect on the market. Dick provided examples to illustrate these concepts, noting that fair use is more likely for transformative, non-commercial, and factual works, and less likely for creative works and large portions of copyrighted material.

Dick discussed two significant copyright cases: Andy Warhol Foundation for Visual Arts v. Goldsmith, and Google LLC v. Oracle. In the Warhol case, the Supreme Court ruled 7-2 in favor of photographer Lynn Goldsmith, determining that Andy Warhol's Orange Prince illustration was not transformative enough to qualify as fair use, despite being used in a different commercial context than the original photograph. In the Google case, Dick provided background on the dispute between Google and Oracle over Java SE copyright, noting that Google's use of approximately 11,500 lines of Java code in Android was at the center of the legal battle, though the full details of the Supreme Court's decision were not covered in this segment.

The Supreme Court ruled that Google's use of Java API declarations in Android was fair use, reversing the lower court's decision. Justice Breyer's majority opinion emphasized that the copied code was functional and part of a user interface, while the dissent by Justices Thomas and Alito argued that the code was derivative rather than transformative. The decision highlights the tension between copyright and patent law in protecting computer code, as the declaring code could potentially be better suited for patent protection rather than copyright.

The discussion focused on copyright law and fair use, particularly regarding Anthropic's settlement with publishers over the use of pirated books. The judge in the case ruled that training AI chat boxes on copyrighted books is not illegal under fair use, but the settlement was reached before a final court decision. The group discussed how open source sharing differs from unauthorized distribution of copyrighted material, and explored examples of how legal precedents can evolve over time, including references to historical cases like Napster and the Robertson bolt patent dispute.

The discussion focused on copyright and AI, particularly regarding the use of pirated content and fair use considerations. Dick explained that AI companies are likely aware of copyright laws and may use the Breyer opinion to argue for fair use in most cases. They also discussed the recent agreement between AI companies and major medical journals like the New England Journal of Medicine and JAMA, allowing AI to access and use their content through “open evidence.” This could lead to a monopoly on information by companies like OpenAI, as they would have exclusive access to copyrighted material. Dick also touched on the impact of AI on content creators and publishers, comparing it to the early days of search engines and the challenges faced by smaller websites.

The meeting focused on copyright and AI-generated artwork. Dick explained that AI-generated works cannot be copyrighted by the AI itself, but human input and creativity are required. He discussed a case where an artist added manual changes to an AI-generated image, which was accepted for copyright. The group explored whether the same prompts could be used across different AI engines to create multiple copyrighted images, with Dick explaining that while the images themselves could be copyrighted, the prompts used to create them could also be protected. The conversation ended with a brief discussion about patenting AI processes and the limitations of using existing tools in creative work.