While innovation often outpaces regulation, organizers of an AI conference at Harvard Law say the unprecedented rate of technological change “makes it even harder for the already trailing legal system to catch up”
Jan 16, 2025
At its core, intellectual property (IP) law is a motivator that rewards the pursuit of new ideas. By giving innovators exclusive rights over their work, IP law provides individuals a valuable incentive to say new things, make new tools, and deploy the power of human ingenuity.
Ironically, innovations in the tools used to express these original or distinctive ideas require continuous updates to the laws governing what kind of work deserves protection. Artificial Intelligence (AI) is just the latest of these innovations to emerge … or is it? Although new tools are nothing new (see: the wheel), the unique power wielded by AI offers both unprecedented opportunities and daunting challenges that call into question what we even mean by “creation.”
This year, Fisher and Lee, along with Harvard Law School Lecturer on Law Louis W. Tompros ’03, focused the IP conference specifically on the intersection of IP and AI. Lee and Tompros met with Harvard Law Today to discuss the conference and the intersection of AI and IP.
William Lee and Louis Tompros
Harvard Law Today: Why was the Artificial Intelligence and Intellectual Property Law Conference in November convened? Why is it important to be talking about AI and IP right now?
William Lee: In the past, this event has been much more focused on the specifics of the law and comparisons of the different approaches across jurisdictions. This year, the conference addressed AI more generally with moderators and panelists from a wider variety of fields including homeland security, life sciences, technological development, non-profit advocacy, and even ethics. I think it was an introduction into AI for many of the people in the room and who better to provide that introduction than [Harvard Law School Professor] Jonathan Zittrain ’95. Matt Ferraro, senior counselor for cybersecurity and emerging technology to the secretary of Homeland Security and executive director of the Artificial Intelligence Safety and Security Board, led a panel primarily of industry leaders, explaining the capabilities and trajectory of AI technology. Then, Iain Cunningham from NVIDIA chaired an excellent panel mostly composed of academics and people from industry discussing how IP law and AI interact. We also had input from experts on the AI and IP relationship in jurisdictions across the globe, including Europe, the UK, and Africa, on a panel moderated by Terry Fisher that was particularly illuminating. Then, we closed with a judges panel where a group of five Federal Circuit and two District Court judges offered views on AI issues as well as IP more broadly.
Louis Tompros: IP law has historically, and inherently, operated at the intersection of law and fast-moving technology. Artificial Intelligence is currently where technology is moving the fastest and where the law has the most ground to cover in order to keep pace. This conference was designed to educate folks about AI technology and the various IP law approaches taken in the United States and around the world, and to help forecast how protections for creative and useful ideas will function in the context of these innovative systems. We try to make the IP conference as broadly appealing and relevant to the groups of constituents that are interested in participating, that is, people within the legal community, the business community, and the academic community, including Harvard Law School students. This year was the first time ever that the conference was fully subscribed via pre-registration which is, I think, a great testament to the level and breadth of interest. You can tell that we got it right precisely because of the incredible interest in this year’s event.
HLT: Throughout history, innovations have compelled IP law to adjust and evolve to account for new technology, like the radio, the television, and the internet. Is AI different?
Tompros: The law can’t possibly stay ahead. It will always lag a bit behind. Honestly, that’s part of the fun of IP law because the law is perpetually having to evolve by necessity to keep pace with rapidly evolving aspects of technology. I don’t think AI is different in kind from previous technological revolutions that affected the law, but I do think it is quite possibly different in scale. The pace of the development of the technology here is so accelerated that the speed at which technological advances are coming makes it even harder for the already trailing legal system to catch up. That leads to some interesting possibilities, but it also leads to some serious challenges. Ultimately, it demands creative and innovative thinking in the design of legal structures established to try to manage it.
“The law can’t possibly stay ahead, it will always lag a bit behind. Honestly, that’s part of the fun of IP law because the law is perpetually having to evolve by necessity to keep pace with rapidly evolving aspects of technology.”
Louis Tompros
Lee: Louis is 100% right. Whatever’s happening across the quad in the engineering or STEM buildings at Harvard will always inevitably provide a new batch of issues for the law school to investigate a decade from now. On the other hand, one area where AI is unique is that it has emerged in a decentralized fashion. The vast majority of advances in science over the last century, such as the advent of molecular biology and genetic engineering, were a product of NIH [National Institutes of Health] funding at institutions like Harvard, MIT, etc. The internet, too, was primarily the product of Department of Defense funding.
AI, on the other hand, has bypassed the traditional network of institutional support and stewardship. It has arisen amid the decentralization of the innovation engine driven by a host of individual players, many of them representing for-profit enterprises. That is why AI has evolved, and will continue to evolve, at a much more rapid pace. That is also why the questions about who is regulating this nascent technology, and associated industries, become critically important.
Lastly, I would like to point out that past advances like genetic engineering, for example, mostly required changes to relevant patent law. What became abundantly clear during this conference, though, is that AI, still only in its infancy, has already demonstrated it will impact patent, trademark, trade dilution, copyright, antitrust, competition and trade, and a myriad of other laws, rights, and regulatory systems. In that sense, these wide-ranging legal implications also make AI significantly harder to understand, unpack, and address.Lastly, I would like to point out that past advances like genetic engineering, for example, mostly required changes to relevant patent law. What became abundantly clear during this conference, though, is that AI, still only in its infancy, has already demonstrated it will impact patent, trademark, trade dilution, copyright, antitrust, competition and trade, and a myriad of other laws, rights, and regulatory systems. In that sense, these wide-ranging legal implications also make AI significantly harder to understand, unpack, and address.
HLT: What aspects of the legal profession do you expect AI to have the greatest impact on going forward?
Lee: It’s everywhere, right? At the moment within any given law firm, the leaders, partners, and associates have to be thinking about how they can use AI. Without a doubt, the legal profession can use AI to become much more efficient. The next question becomes, “How can you use it to deliver services more efficiently, more effectively, and better to your clients?” The third question is, “What guardrails and guidelines can be established to ensure compliance with the relevant standards for administrating justice?” The fourth question is, “How do you do this while safeguarding the security of confidential information and your own internal innovations?” Then, a fifth is, “Who will have access to this marketplace? The usual suspects, or someone in a little shop around the corner in Kendall Square?”
“Without a doubt, the legal profession can use AI to become much more efficient.”
Bill Lee
Tompros: I agree completely. To take it to the philosophical level, AI is fundamentally about pattern recognition. So much of what we do in law, at its core, involves recognizing and applying patterns from what happened before in order to accurately predict and decide what will happen in the future. In that sense, AI is incredibly relevant to the kind of work firms handle today, which is the way our legal system has worked for quite a long time.
HLT: What aspects of AI and IP law do you expect to change the most going forward? What aspects, if any, will have the biggest impact on litigation?
Tompros: Unsurprisingly, the first wave of litigation that we’re seeing in the AI/IP space relates to copyright because, fundamentally, AI is about teaching machines how to recognize patterns and execute on them. That requires machines to learn, so AI training involves copying the information coming in. That’s mostly a good thing because it translates to better AI. We want machines to understand how to do things well enough to be able to provide benefits. There are currently 20 to 30 cases working their way through the courts attempting to draw the line between fair use and copyright infringement in the context of AI learning. I think that the guardrails set by these pending cases in the U.S. and European Union will be very instructive for the rest of the world going forward.
I think the next wave to come will be about what happens in the IP space when you begin to apply AI to the work of businesses generating and creating. So, what implications will there be for the patent system when you have artificial intelligence doing something in a way ordinary inventors may, or may not, be able to do themselves? What is going to be protected? What rights will individuals and companies have when it comes to inventing things in the context of creative computer technologies? The same questions will likely be relevant for the cybersecurity space as well. At the moment, we are largely focused on training AI: What is allowed, what will be permitted, and how people will be compensated. Going forward, though, I think debates will unfold over degrees of accommodation for artificial intelligence in the context of inventions, security, and privacy.
Lee: It’s virtually impossible to predict. A recent publication by a Harvard scholar named Andrew Hartwig argues that, over the next decade or so, 90% of the IP generated nationally and globally is probably going to be generated by Artificial Intelligence. I think that if you take Professor Tompros’s very precise, correct summary and then view it in the context that AI-generated content may occupy 90% of the space of innovation, you start to understand the gravity and the magnitude of the impact that’s coming. One of the discussions at the conference involved a proprietary algorithm with the sole function of taking all the publicly available information on the best possible configurations of therapeutics, and combining it with privately generated information, to design perfect therapeutics and trials. Essentially, something that used to take teams of organic chemists a decade to design, test, and release can now happen in a year or two.
HLT: What was the most surprising discovery, innovation, or aspect of the AP/IP crossover you learned about at the November conference?
Lee: One thing that stood out to me from the conference was the fact that, when the internet emerged about 25 years ago, there were a lot of advocates who vigorously opposed regulating it. I think our audience was surprised to learn that those same advocates, today, are some of the biggest proponents of regulating AI. I think another important revelation to many of the attendees was how AI transcends the geographic barriers that often define jurisdictions and the respective policies that apply. Due to the decentralized nature of AI’s proliferation, IP laws and regulations may be of no significance at all if the information produced by AI can be generated in one jurisdiction and used in another. In one of the conference panels, we heard about the ongoing trend of considerable venture capital investments directed at AI in Africa, specifically, in the countries least likely to regulate. To me, that was the single biggest revelation and another salient point in support of considering international collaboration on AI/IP standards.
Tompros: While I wasn’t exactly surprised by the quality of the dialogue at the conference, the level of critical thinking that took place was genuinely heartening. I can imagine people going into this conference thinking, “This is going to be a conversation we’ve heard a million times. On one side, you have a whole bunch of academics and government officials advocating for strict standards and regulations. On the other side, you have a whole bunch of folks from the private sector and technology advocating for no regulation.”
Interestingly enough, that really wasn’t the case. We saw a very honest and robust discussion at the conference between experts from various backgrounds investigating the best ways to manage these issues. Across industry, government, and academia, the panelists at this conference were clearly prepared to ask hard, thoughtful questions. That might not have been a total surprise, but it was certainly reassuring.Interestingly enough, that really wasn’t the case. We saw a very honest and robust discussion at the conference between experts from various backgrounds investigating the best ways to manage these issues. Government officials were discussing appropriate uses of AI and sharing their ideas for public sector applications. Folks from the private industry side were talking about ethical standards and evaluating how to manage AI use through regulations in a just and appropriate manner. Across industry, government, and academia, the panelists at this conference were clearly prepared to ask hard, thoughtful questions. That might not have been a total surprise, but it was certainly reassuring.
HLT: From a policymaking standpoint, were there any specific gaps at the national or global level that this conference may have helped identify a need for?
Lee: I think that the folks who were on the panels are people who recognize all the risks and benefits of AI and the ambiguity of its horizon. They’re deeply focused on the different ramifications and consequences of AI and its rapid development. There was discussion about how the foundation world needs to ensure that AI does not become another tool, unintentionally, to increase the divide between the very wealthy and the very poor. It could, very easily. Our conferences are never intended to be overly prescriptive or culminate in a game plan with specific steps. The goals are more about engaging people from different venues, hearing each other, and then going back to their respective venues with newly informed perspectives.
Tompros: I think Bill summarized it exactly right. If our overall goal was that people think about these issues with a greater degree of depth and sophistication than they had when they came in, I think we certainly accomplished that.
Want to stay up to date with Harvard Law Today? Sign up for our weekly newsletter.
