Three Laws Blog: Law and Artificial Intelligence

Strong on Artificial Intelligence in Civil Justice Systems: An Empirical and Interdisciplinary Analysis and Proposal for Moving Forward

S. I. Strong, Emory University School of Law, is publishing Artificial Intelligence in Civil Justice Systems: An Empirical and Interdisciplinary Analysis and Proposal for Moving Forward in volume 41 of the Ohio State Journal of Dispute Resolution. Here is the abstract.

Artificial intelligence – particularly generative artificial intelligence – has become one of the leading challenges of our time. One particular area of concern involves the effect of generative artificial intelligence on civil justice systems, meaning both litigation and arbitration. This Article undertakes an innovative interdisciplinary analysis of the effect of artificial intelligence on the legal system and the legal profession. Incorporating empirical research from the social sciences, the discussion identifies various individualistic problems affecting judges, arbitrators, lawyers, and law students, while also considering systemic harms to civil justice writ large. Identifying problems is good, but identifying solutions is better. This Article therefore includes a novel means of balancing the benefits of artificial intelligence while simultaneously protecting against its more pernicious elements. This proposal – which looks to England for inspiration – discusses how legal education and the legal profession might be structured in a post-AI world to allow for the appropriate use of generative AI by judges, arbitrators, and lawyers.

Download the article from SSRN at the link.

Lee on Beyond Algorithmic Disgorgement: Remedying Algorithmic Harms

Christina Lee, George Washington University Law School, is publishing Beyond Algorithmic Disgorgement: Remedying Algorithmic Harms in volume 16 of the U.C. Irvine Law Review. Here is the abstract.

AI regulations are popping up around the world, and they mostly involve ex-ante risk assessment and mitigating those risks. But even with careful risk assessment, harms inevitably occur. This leads to algorithmic remedies: what to do once algorithmic harms occur, especially when traditional remedies are ineffective. What makes a particular algorithmic remedy appropriate for a given algorithmic harm? I explore this question through case study of a prominent algorithmic remedy: algorithmic disgorgement—destruction of models tainted by illegality. Since the FTC first used it in 2019, it has garnered significant attention, and other enforcers and litigants around the country and the world have started to invoke it. Alongside its increasing popularity came a significant expansion in scope. Initially, the FTC invoked it in cases where data was allegedly collected unlawfully and ordered deletion of models created using such data. The remedy’s scope has since expanded; regulators and litigants now invoke it against AI whose use, not creation, causes harm. It has become a remedy many turn to for all things algorithmic. I examine this remedy with a critical eye, concluding that though it looms large, it is often inappropriate. Algorithmic disgorgement has evolved into two distinct remedies. Data-based algorithmic disgorgement seeks to remedy harms committed during a model’s creation; use-based algorithmic disgorgement seeks to remedy harms caused by a model’s use. These two remedies aim to vindicate different principles underlying traditional remedies: data-based algorithmic disgorgement follows the disgorgement principle underlying remedies like monetary disgorgement and the exclusionary rule, while use-based algorithmic disgorgement follows the consumer protection principle underlying remedies like product recall. However, they often fail to live up to the principles. AI systems exist in the context of the algorithmic supply chain; they are controlled by many hands, and seemingly unrelated entities are connected to each other in complicated ways through complex data flows. The realities of algorithmic supply chain means that algorithmic disgorgement is often a bad fit for the harm at issue and causes undesirable effects throughout the algorithmic supply chain, imposing burden on innocent parties while not imposing cost on the blameworthy; ultimately, algorithmic disgorgement undermines the principles it seeks to promote. From this analysis, I derive considerations for determining whether an algorithmic remedy is appropriate—the responsiveness of the remedy to the harm and the full impact of the remedy throughout the supply chain—and underscore the need for a diversity of algorithmic remedies.

Download the article from SSRN at the link.

Galler and Soled on AI and the Regulation of Tax Preparers

Linda Galler, Hofstra University School of Law, and Jay A. Soled, Rutgers University, are publishing AI and the Regulation of Tax Preparers as a Hofstra Univ. Legal Studies Research Paper (forthcoming). Here is the abstract.

Tax return preparers play a pivotal role for millions of taxpayers, and their number ranges in the hundreds of thousands. Despite preparers’ importance in the tax return submission process and their numerical size, they are largely unregulated. Because of this lack of oversight, a significant number of tax return preparers put their own financial interests and those of their clients ahead of the government’s financial interests. In the past, there have been calls for increased regulatory oversight to keep wayward practitioners’ actions in check. Therefore, a decade and a half ago, Congress, in conjunction with the Treasury Department, mandated that every paid tax return preparer secure a Preparer Tax Identification Number (PTIN) and affix it to the face of the tax returns that they prepare. Enter artificial intelligence (AI)—with its unparalleled ability to process and analyze an immense amount of data—combined with electronic tax return filings. With immediate access to PTINs and other information or characteristics associated with such returns, AI is ideally situated to identify tax return preparers who are derelict in their professional duties. Once identified, the IRS can hold these practitioners accountable for their mistakes and misdeeds. Unlike the prior academic papers that have championed regulatory oversight well beyond PTIN registration (e.g., mandatory annual education courses and exams), this analysis stands apart. Instead, it recognizes AI’s power and what it can accomplish if properly brought to bear to the problem of unregulated tax return preparers. Seizing this opportunity would most likely result in increased tax compliance, the collection of billions more in revenue without raising taxes, and a narrowing of the tax gap (i.e., the difference between what taxpayers pay in tax and what they owe). Furthermore, the foregoing may be accomplished without adding needless administrative burdens.

Download the article from SSRN at the link.

Filippi, Tomlinson, and Torrance on The Law and AI as "Apex Collaborator": Legal Frameworks for Optimized Cooperation

David S. Filippi, Western University of Health Sciences, Bill Tomlinson, University of California, Irvine; Victoria University of Wellington, Re Herenga Waka, and Andrew W. Torrance, University of Kansas School of Law; MIT Sloan School of Management, are publishing The Law and AI as "Apex Collaborator": Legal Frameworks for Optimized Cooperation in the FIU Law Review. Here is the abstract.

Law fundamentally exists to enable human cooperation, providing frameworks for everything from basic contracts to complex international agreements. As artificial intelligence systems grow more sophisticated, they may enable new ways that collaborative activity can occur. We posit the possibility of a new kind of AI entity: the “Apex Collaborator”, a computational system with capabilities for cooperation and partnership that are superior, in at least some ways, to those of humans. Just as apex predators shape the ecosystems in which they live through predation, Apex Collaborators would shape human-AI networks through their ability to enhance peaceful coexistence, collective problem-solving, and shared decision-making. “AI as an Apex Collaborator” flips the normal scripts of “AI as danger” or “AI as passive deliverer of benefits to humans”, instead conceiving of AI as a catalyst and enabler capable of lifting human abilities to cooperate above their evolutionary trajectory. This article maps the legal architecture needed to guide AI development toward this collaborative potential, while simultaneously mapping fundamental implications particular coding decisions may have for the law. We address key areas requiring reform: liability regimes governing potential harms to humans, property, or other AIs, copyright law to enable AI training, structures and strictures for AI self-determination, clear accountability for AI-assisted actions and AI agents, interoperability standards, and alignment requirements. The article proposes specific proactive and enforcement mechanisms for AI-ogenic conflict resolution, military restrictions, and data protection including cross-border transfer controls. We outline pathways to foster beneficial collaboration while preventing harmful applications. In particular, we explore the potential for AIs acting as Apex Collaborators to support humanity’s transition to sustainability. Our framework recognizes that as AI systems advance toward apex collaboration capabilities, they may need to participate in their own governance, monitoring and responding to not only harmful AI developments but also previously impossible benefits to humanity.

Shereshevsky and Shany on Programmed to Obey: The Limits of Law and the Debate over Meaningful Human Control of Autonomous Weapons

Yahli Shereshevsky, University of Haifa Faculty of Law, and Yuval Shany, Hebrew University of Jerusalem, Faculty of Law and Institute of Criminology; Israel Democracy Institute, are publishing Programmed to Obey: The Limits of Law and the Debate over Meaningful Human Control of Autonomous Weapons in volume 57 of the Columbia Human Rights Law Review. Here is the abstract.

New military technologies are transforming the contemporary battlefield and raise complex ethical and legal questions previously unaddressed. This essay makes three novel contributions to the significant debate on Autonomous Weapon Systems (AWS) and military AI in the legal and ethical literature. First, it puts forward a normative argument against AWS—even if they outperform humans in adhering to the rules governing the conduct of hostilities. This argument is grounded in the critical importance of the human capacity to act beyond the strict letter of the law. The essay contends that this capacity is central to the regulation of warfare, which permits, rather than obligates, the use of force against legitimate targets. Second, it offers a doctrinal analysis of International Humanitarian Law (IHL) and International Human Rights Law (IHRL)—the two principal legal regimes that regulate armed conflicts under international law—providing a fresh perspective on how they intersect in the context of AWS. Finally, the essay explores the extent to which its normative argument is persuasive in the context of military AI beyond AWS, an area that is rapidly evolving and already extensively employed in current conflicts. It examines the similarities and differences between these emerging technologies, and reflects on the implications of those differences for the desirable regulation of both technologies.

Download the article from SSRN at the link.

Campagnolo, van Haernen, Ravitharan, and Murray on Artificial Intelligence's Impact on Legal Journals

Yan Campagnolo, Leonie van Haernen, Shaarini Ravitharan, and Emma Murray, all of the University of Ottawa, Common Law Section, have published Artificial Intelligence’s Impact on Legal Journals as Ottawa Faculty of Law Working Paper No. 2025-04. Here is the abstract.

This report provides an overview of the opportunities and challenges presented by the use of artificial intelligence (AI) and its impact on legal journals, including the Ottawa Law Review (OLR). Throughout the publication lifecycle of a given piece, AI tools may play a pivotal role in enhancing the editorial and publishing processes. Similarly, authors submitting to legal journals may also leverage AI tools for purposes that range from improving readability to generating content. While the potential benefits are significant, the use of such tools raises various issues pertaining to the accuracy and quality of publications, as well as broader ethical and legal issues. Journals—both legal and non-legal—have responded to these opportunities and challenges at different speeds and in different ways. Some journals in non-legal disciplines have developed extensive AI policies, while the majority of legal journals—particularly in Canada—appear to be falling behind in this regard. This report contains several recommendations that will empower the OLR to embrace the transformative potential of AI responsibly while maintaining its commitment to safeguarding privacy, intellectual property, and scholarly rigour. Central to this endeavour is the adoption of three AI policies: one covering the use of generative AI and AI-assisted technologies in submissions, another addressing AI usage in the peer review process, and a final one relating to the editorial team. This report ultimately aspires to ensure the OLR upholds its reputation as a reliable contributor to legal scholarship by guiding the organization through this new technological revolution.

Download the report from SSRN at the link.

Campagnolo, van Haernen, Ravitharan, and Murray on Artificial Intelligence's Impact on Legal Journals

Yan Campagnolo, Leonie van Haernen, Shaarini Ravitharan, and Emma Murray, all of the University of Ottawa, Common Law Section, have published Artificial Intelligence’s Impact on Legal Journals as Ottawa Faculty of Law Working Paper No. 2025-04. Here is the abstract.

This report provides an overview of the opportunities and challenges presented by the use of artificial intelligence (AI) and its impact on legal journals, including the Ottawa Law Review (OLR). Throughout the publication lifecycle of a given piece, AI tools may play a pivotal role in enhancing the editorial and publishing processes. Similarly, authors submitting to legal journals may also leverage AI tools for purposes that range from improving readability to generating content. While the potential benefits are significant, the use of such tools raises various issues pertaining to the accuracy and quality of publications, as well as broader ethical and legal issues. Journals—both legal and non-legal—have responded to these opportunities and challenges at different speeds and in different ways. Some journals in non-legal disciplines have developed extensive AI policies, while the majority of legal journals—particularly in Canada—appear to be falling behind in this regard. This report contains several recommendations that will empower the OLR to embrace the transformative potential of AI responsibly while maintaining its commitment to safeguarding privacy, intellectual property, and scholarly rigour. Central to this endeavour is the adoption of three AI policies: one covering the use of generative AI and AI-assisted technologies in submissions, another addressing AI usage in the peer review process, and a final one relating to the editorial team. This report ultimately aspires to ensure the OLR upholds its reputation as a reliable contributor to legal scholarship by guiding the organization through this new technological revolution.

Download the report from SSRN at the link.

Galler and Soled on AI and the Regulation of Tax Preparers

Linda Galler, Hofstra University School of Law, and Jay A. Soled, Rutgers University, are publishing AI and the Regulation of Tax Preparers as a Hofstra Univ. Legal Studies Research Paper (forthcoming). Here is the abstract.

Tax return preparers play a pivotal role for millions of taxpayers, and their number ranges in the hundreds of thousands. Despite preparers’ importance in the tax return submission process and their numerical size, they are largely unregulated. Because of this lack of oversight, a significant number of tax return preparers put their own financial interests and those of their clients ahead of the government’s financial interests. In the past, there have been calls for increased regulatory oversight to keep wayward practitioners’ actions in check. Therefore, a decade and a half ago, Congress, in conjunction with the Treasury Department, mandated that every paid tax return preparer secure a Preparer Tax Identification Number (PTIN) and affix it to the face of the tax returns that they prepare. Enter artificial intelligence (AI)—with its unparalleled ability to process and analyze an immense amount of data—combined with electronic tax return filings. With immediate access to PTINs and other information or characteristics associated with such returns, AI is ideally situated to identify tax return preparers who are derelict in their professional duties. Once identified, the IRS can hold these practitioners accountable for their mistakes and misdeeds. Unlike the prior academic papers that have championed regulatory oversight well beyond PTIN registration (e.g., mandatory annual education courses and exams), this analysis stands apart. Instead, it recognizes AI’s power and what it can accomplish if properly brought to bear to the problem of unregulated tax return preparers. Seizing this opportunity would most likely result in increased tax compliance, the collection of billions more in revenue without raising taxes, and a narrowing of the tax gap (i.e., the difference between what taxpayers pay in tax and what they owe). Furthermore, the foregoing may be accomplished without adding needless administrative burdens.

Download the article from SSRN at the link.

Daly on Artificial Intelligence and Administrative Tribunals

Paul Daly, University of Ottawa, Common Law Section, has published Artificial Intelligence and Administrative Tribunals. Here is the abstract.

In this paper, I will examine the implications of artificial intelligence for administrative tribunals. I will briefly set the scene in Part I by describing the potential advantages and disadvantages of advanced technology for administrative tribunals. Although I have reservations about integrating AI into the daily operations of public administration, administrative tribunals may be a uniquely favoured site for technological experimentation. Since their origins in the 19th century the raison d’etre of administrative tribunals has been more efficient and effective decision-making. The very thing that artificial intelligence promises – efficiency and effectiveness – is exactly that which administrative tribunals have long primarily been designed to achieve. This is not to suggest, however, that administrative tribunals should move blindly forward and embrace AI in all shapes and forms. In Part I, I also suggest that careful, contextual analysis is required before using AI and lay out several typologies that are helpful in conducting the necessary analysis. In Part II, I then discuss the ways in which administrative tribunals can use AI: communicating with the public and enhancing the accessibility of their websites; assigning decisions more effectively to members of the tribunal; triaging files to ensure decisions are made as quickly as possible; drafting decisions; and, conceivably, making decisions. These use cases are set out in increasing order of audacity. Machines making websites more usable for individuals is one thing, making life-defining decisions about individual rights and obligations quite another. In the end, though, I will argue that there is a broad range of potential uses for AI in administrative tribunals.

Download the paper from SSRN at the link.

Ye, Luceri, and Ferrara on Auditing Political Exposure Bias: Algorithmic Amplification on Twitter/X Approaching the 2024 U.S. Presidential Election

Jinyi Ye, Luca Luceri, and emilio Ferrara, all of the University of Southern California, have published Auditing Political Exposure Bias: Algorithmic Amplification on Twitter/X Approaching the 2024 U.S. Presidential Election. Here is the abstract.

Approximately 50% of tweets in X's user timelines are personalized recommendations from accounts they do not follow. This raises a critical question: what political content are users exposed to beyond their established networks, and how might this influence democratic discourse online? Due to the black-box nature and constant evolution of social media algorithms, much remains unknown about this aspect of users' content exposure, particularly as it pertains to potential biases in algorithmic curation. Prior research has shown that certain political groups and media sources are amplified within users' in-network tweets. However, the extent to which this amplification affects out-of-network recommendations remains unclear. As the 2024 U.S. Election approaches, addressing this question is essential for understanding the influence of algorithms on online political content consumption and its potential impact on users' perspectives. In this paper, we conduct a three-week audit of X's algorithmic content recommendations using a set of 120 sock-puppet monitoring accounts that capture tweets in their personalized "For You" timelines. Our objective is to quantify out-of-network content exposure for right-and left-leaning user profiles and to assess any potential biases in political exposure. Our findings indicate that X's algorithm skews exposure toward a few high-popularity accounts across all users, with rightleaning users experiencing the highest level of exposure inequality. Both left-and right-leaning users encounter amplified exposure to accounts aligned with their own political views and reduced exposure to opposing viewpoints. Additionally, we observe a right-leaning bias in exposure for new accounts within their default timelines. We hope this empirical study promotes discussion on the transparency and accountability of social media algorithms, contributing to critical issues on safeguarding election integrity and fostering a more informed digital public sphere.

Download the article from SSRN at the link.