12 October 2024

Nature

'Rights for nature or protecting people’s rights?: The operationalization of rights of nature in non-indigenous communities in the United States' by Ellen Kohl in (2024) 156 Geoforum comments 

 As the Rights of Nature movement gains political traction globally, researchers must examine how this transnational movement to extend rights to nature or natural entities is being operationalized in place. Proponents of the rights of nature contend transferring rights to nature constitutes a paradigm shift in human-environment interactions and will lead to solutions directed at the root causes of environmental problems. Critics contend that these rights-based governance structures have the potential to do more harm than good for environmental protection depending on the cultural and legal frameworks within which rights of nature are enacted. In this paper, I examine how rights of nature have been operationalized in non-Indigenous communities in the United States through an analysis of rights of nature ordinances passed in these communities between 2006 and 2020. Drawing on theoretical engagements with rights I demonstrate how the reliance on universalizing human rights frameworks and anti-corporation rhetoric both distinguish these ordinances from the broader rights of nature movement and center the rights of people to have access to a clean environment rather than the intrinsic rights of nature. In conclusion, I explore alternatives to how rights of nature are currently operationalized in non-Indigenous communities in the United States and call for increased research on the implications and impacts of rights of nature ordinances to assess whether they achieve their stated goals. 

Rights of Nature (RoN) is a transnational “rights-based environmental governance” structure that uses liberal legal rights to protect nature rather than relying on existing forms of environmental governance (Kohl and Walenta 2023). The movement traces its intellectual origins to Earth Jurisprudence scholarship which contends that current environmental crises are caused by cultural and legal systems in which human needs are separate and superior to those of nature (Berry, 1999, Boyd, 2017, Cullinan, 2011). The RoN movement draws on different legal channels to grant nature legal rights akin to the rights secured by human persons, such as the right to live, thrive and flourish, and the right to not be polluted (Gilbert et al., 2023, Kauffman and Martin, 2021, Kauffman and Martin, 2018; O’Donnell, 2023, O’Donnell, 2020, Putzer et al., 2022, Tănăsescu, 2022). This can be done by granting personhood status to nature, either directly or indirectly through the appointment of an administrating body, or by extending rights to nature (Kauffman and Martin, 2021, Kurki, 2022, Tănăsescu, 2020). RoN advocates contend this structure provides greater protection to natural entities and will prevent catastrophic environmental destruction (Gilbert et al., 2023, O’Donnell, 2023). For many, the extension of rights to nature represents a paradigm shift in the human-environment relationship (Cuadros, 2019, Dogaru and Dogaru, 2022a, Dogaru and Dogaru, 2022b, Ito and Montini, 2018, Villavicencio-Calzadilla and Kotzé, 2023). As more communities turn to RoN activism, and adopt RoN legal frameworks, there is an emerging theoretical debate in support of (Bradshaw, 2022, Kauffman and Martin, 2023) and those raising concerns about extending rights to nature (Bellina, 2024, Guim and Livermore, 2021, Sachs, 2023, Spitz and Penalver, 2021, Petel, 2024). 

The motivations for transferring rights to nature vary over space and time but O’Donnell, writing in Eckstein et al. (2019), contends there at least four specific reasons that rivers, the natural entity she studies, are granted rights. These include (1) a recognition of Indigenous cosmologies and values, (2) eco-centrism, or bringing rivers and humans onto equal level, (3) market environmentalism, creating mechanisms through which rivers can participate in markets, and (4) private interest regulatory theory, which allows mechanisms through which rivers can advocate for themselves (Eckstein et al. 2019). In this paper, I add a motivation of people’s desires to address a particular environmental concern for which traditional forms of environmental rule making are not providing them sufficient protection. In this way, I frame the understandings of RoN not only through legal understandings, but also in place and through political understandings because “one cannot understand what the rights of nature are doing without thinking about them in terms of power relations” (Tănăsescu 2022, 16). Given the diversity of how, why, and where RoN are adopted the operationalizations of RoN looks different in different places. While RoN is a transnational movement, the specific place-based articulations of the movement and the outcomes of RoN activism are impacted by the social, cultural, and political context within which they take place (Kinkaid, 2019, Gilbert et al., 2023, Tănăsescu et al., 2024). Through an examination of the over 60 RoN ordinances enacted within the United States (US) between 2008 and 2020 I build on current scholarship examining the operationalization of RoN in the US context (Fitz-Henry, 2018, Fitz-Henry, 2022, Fitz-Henry, 2023, Kohl and Walenta, 2023; Macpherson, 2021, Moutrie, 2022) to expand our understandings of the place-based articulations of RoN. 

In this paper, I carry out this analysis to examine how theoretical conceptions of RoN have been translated in US legal structures. Through this analysis and discussion, I demonstrate that the operationalization of RoN in non-Indigenous communities in the US shifts from granting nature the status of personhood to an extension of rights to nature by extending conceptions of human rights onto nature. I demonstrate how this universalization centers the needs of people rather than the intrinsic rights of nature. Moreover, the operationalization of RoN in these communities specifically has an anti-corporation clause. These differences contradict the claims of RoN proponents who justify the need for RoN legislation as necessary because it represents a paradigm shift by elevating nature’s rights as equal to people’s rights. I begin by reviewing conceptions of property, rights, personhood, and rights to nature, paying particular attention to the US context. From there, I explain the methods used. Then I use the theoretical frameworks to analyze the ordinances enacted in the US context. In conclusion, I call for more research on how, if at all, these ordinances are changing people’s perceptions of nature and if environmental conditions have improved in and around places where RoN legislation has been enacted.

'Civil Litigation for Environmental Damages: Are the So-Called Rights of Nature an Alternative?' by Guillermo Schumann Barragan in (2024) 14 International Journal of Procedural Law 73–98 comments 

This paper aims to analyse civil litigation for environmental damages and whether the so-called rights of nature are a viable alternative de lege ferenda. To this end, it examines (i) the participation of civil society in environmental protection through public enforcement mechanisms in the Spanish legal order; (ii) the cause of action for environmental damage claims; and (iii) whether rights of nature are a good option to enhance the protection of nature in civil justice. 

On the 3rd of October 2022, Law 19/2022 on the recognition of the legal status of the Mar Menor lagoon was published in the Spanish Official State Gazette (boe). The law attributes the lagoon legal status (Article 1 Law 19/2022) and certain rights: the right to exist, protection, conservation and restoration (Article 2 Law 19/2022). Furthermore, according to Article 6, “any natural or legal person has the right to defend the ecosystem of the Mar Menor”. Depending on the nature of the action, criminal, civil or administrative courts will have subject-matter jurisdiction (Article 4 Law 19/2022). 

The attribution of legal status to elements of nature – e.g. a river, a forest, a mountain or a coral reef – is not new on the international scene. This theory, which is unknown in European legal orders, is known as the rights of nature. As a result of public concern about climate change, environmental protection has become one of the EU’s political priorities. A good example of this is the European Commission’s package of measures known as the Green Deal. Renewable energy and environmental have become a leitmotiv on the European political agenda. 

In this context, this paper aims to analyse civil litigation for environmental damages and whether the so-called rights of nature are a viable alternative de lege ferenda. To this end, it examines (i) the participation of civil society in environmental protection through public enforcement mechanisms in the Spanish legal order; (ii) the cause of action for environmental damage claims; and (iii) whether rights of nature are a good option to enhance the protection of nature in civil justice.

Hegemonies

'Gramsci, Hegemony, and the Law' by Douglas Litowitz in (2000) 2 BYU Law Review comments 

Hegemony is a Marxist concept derived largely from the work of Antonio Gramsci. It emerged as a central theme during the hey-day of the Critical Legal Studies movement, and it remains popular in contemporary legal studies, albeit within a somewhat narrow circle of law-and-society scholars. The concept of hegemony deserves broader consideration from the legal academy because it is a critical tool that generates profound insights about the law’s ability to induce submission to a dominant worldview. The purpose of this article is to introduce Gramsci’s work to a wider audience by explaining, critiquing, and revitalizing his notion of hegemony as it applies to law. 

This article is not merely a description of Gramsci’s influence. I also want to take issue with the direction taken in recent legal scholarship on hegemony. Gramsci spoke about hegemony in the singular as a large-scale national phenomenon (e.g., the hegemony of a single dominant group over all others), and that is what made his theory powerful—he was describing a phenomenon that permeated all of our lives. Recent legal scholarship eschews Gramsci’s notion of an overarching hegemony in favor of the idea that hegemony occurs only at discrete and disconnected sites such as race, age, disability, and gender. The statement of a leading scholar is instructive of this new approach: “Instead of an overarching hegemony, there are hegemonies. . . . Law cannot be view as hegemonic or not as a whole.” Much of the recent legal scholarship on hegemony builds on this post-Gramscian approach to hegemonies, which is noteworthy for its reliance on historical studies of hegemony and cross-cultural examples of domination and resistance. Many of these scholars have tacitly abandoned the search for an overarching hegemony at work in the current legal system. 

Against this line of scholarship, I will argue in favor of the continuing relevance of Gramsci’s notion of hegemony in the singular. In particular, I will argue that the current legal system is hegemonic in the Gramscian sense in that it induces people to comply with a dominant set of practices and institutions without the threat of physical force and that this hegemony is overarching because it encompasses people of diverse races, classes, and genders. The law induces passive compliance in large measure through its function as constitutive of social ontology—it provides rules for the proper construction of authorized institutions and approved activities, such as setting up corporations, buying and selling real estate, drafting wills, hiring employees, and so on; it is a hegemonic code that replicates the social ontology in much the same way that a genetic code replicates a biological organism. 

Part II of this article focuses on Gramsci’s use of the term hegemony and the concept’s implications for legal doctrine and prac-tice. Part III traces the impact of hegemony as a critical concept in legal scholarship over the last twenty years and then proceeds to a sustained critique of recent scholarship on hegemony. The final part, Part IV, draws from literary works to support my reformulation of hegemony as the dissemination of a dominant code composed largely of unchallenged background assumptions that undergird the law. Part IV presents a revised conceptualization of hegemony that does away with Gramsci’s notion that law is the hegemonic tool of a dominant class in favor of the notion that law represents a dominant code or map that perpetuates the status quo and its attendant inequalities, oppressions, and disaffections. This reformulation of Gramsci’s concept of hegemony captures fresh insights about the law’s ability to induce submission and paralysis while avoiding Gramsci’s reliance on orthodox Marxist categories that are no longer tenable. Still, hegemony remains a critical and negating tool, not a positive concept. That is, the recognition of hegemony is a tool to raise one’s consciousness: it clears away the distortions and artificial boundaries that insulate the existing legal framework, but it cannot provide a blueprint for a better system. This means that Gramsci’s work provides important insights for understanding how the law sustains un-equal power relations, but it offers scant direction for reforming the law

Fakes

The interim report of the Australian Senate's Select Committee on Adopting Artificial Intelligence feastures the following recommendations 

 Recommendation 1 The committee recommends that, ahead of the next federal election, the government implement voluntary codes relating to watermarking and credentialling of AI-generated content. 

Recommendation 2  The committee recommends that the Australian Government undertake a thorough review of potential regulatory responses to AI-generated political or electoral deepfake content, including mandatory codes applying to the developers of AI models and publishers including social media platforms, and prohibitions on the production or dissemination of political deepfake content during election periods, for legislative response prior to the election of the 49th Parliament of Australia. 

Recommendation 3  The committee recommends that laws restricting the production or dissemination of AI-generated political or electoral material be designed to complement rather than conflict with the mandatory guardrails for AI in high-risk settings, the recently introduced disinformation and misinformation reforms, and foreshadowed reforms to truth in political advertising. 

Recommendation 4  The committee recommends that the Australian Government ensure that the mandatory guardrails for AI in high-risk settings also apply to AI systems used in an electoral or political setting. 

Recommendation 5  The committee recommends that the government examine mechanisms, including education initiatives, to improve AI literacy for Australians, including parliamentarians and government agencies, to ensure Australians have the knowledge and skills needed to navigate the rapidly evolving AI landscape, particularly in an electoral context.

11 October 2024

Biosimilars Regulation

'Biosimilar approval pathways: comparing the roles of five medicines regulators' by Ryan P Knox, Vineet Desai and Ameet Sarpatwari in (2024) 11(2) Journal of Law and the Biosciences comments 

Biologics are playing an increasingly important role in health care globally but are placing a substantial burden on payers. The development of biosimilars—drugs that are highly similar to and have no clinically meaningful differences from originator biologics—is critical to improving the affordability and accessibility of these medications. Medicines regulators, however, have had varied success with biosimilars to date. We examined agency guidance documents, peer-reviewed articles, and gray literature related to biosimilars in Australia, Canada, the European Union, the United Kingdom, and the United States to evaluate variations in the approaches to biosimilar approval taken by their respective medicines regulators. We found that the medicines regulators take similar approaches to biosimilar approvals, but that differences in their policies and their jurisdiction’s laws regarding testing requirements, indication extrapolation, exclusivities, and substitution may contribute to the varied successes of biosimilars observed. Policies supportive of product-specific guidance, extrapolation, shorter exclusivity periods, and substitution were correlated with greater success in biosimilar approval and uptake. As medicines regulators work to promote biosimilars, understanding the impact of these laws and policies is crucial. Reforms consistent with these policies can create regulatory environments more supportive of biosimilar approvals, promoting access to affordable biologics for patients globally. 

Biologic drugs are complex biopharmaceuticals manufactured from living organisms or derived from their cells and tissues. They are playing an increasingly important role in medical treatments, revolutionizing care for patients with various cancers, autoimmune conditions, and genetic disorders. At the same time, biologics are taking up a larger share of health care spending. Although biologics—such as high-profile blockbuster medications including Humira (adalimumab), Enbrel (etanercept), and Remicade (infliximab) —comprised only 0.4% of drugs prescribed in the United States (US) in 2018 and only 2% of prescription drug claims in Canada in 2021, they represented 46% of net drug spending and 29% of public drug spending in those countries in those years, respectively. Biologics also account for a similarly high percent of prescription drug spending in Australia, the European Union (EU), and the United Kingdom (UK). This disproportionate spending is due to the high prices of biologics. For example, the list price for the blockbuster immunotherapy Keytruda (pembrolizumab) in 2021 was ~$175,000 per patient per year in the US. By 2022, at least three biologics had US list prices greater than $1 million per patient per year. 

To address the high cost of biologics, governments in several countries have developed specialized pathways for medicines regulators to approve highly similar follow-on biologics that have no clinically meaningful differences from an ‘originator’ (or reference) biologic. These ‘biosimilars’ are intended to promote competition, thereby lowering the price of treatment for patients and health systems ... 

To date, the number of biosimilars that have been approved by medicines regulators has varied widely (Table 1). As of May 31, 2024, the US Food and Drug Administration (FDA) had approved 53 biosimilars, the European Medicines Agency (EMA) 101, UK’s Medicines and Healthcare products Regulatory Agency (MHRA) 93, Health Canada 58, and Australia’s Therapeutic Goods Administration (TGA) 55. Although EMA has been approving biosimilars longer than the other regulators—since 2006—only in the last 5 years have other regulators had similar approval rates

10 October 2024

No, Just No

In Dayal [2024] FedCFamC2F 1166 Humphreys J considered an solicitor who used AI in generating a list and summary of authorities that, oops but unsurprising given AI hallucinations, were later  acknowledged by the solicitor not to exist. The accuracy of the document was not verified by the solicitor. 

 The judgment states 

This matter relates to my decision to refer the conduct of a solicitor to the Office of the Victorian Legal Services Board and Commissioner. The solicitor in question tendered to the court a list and summary of legal authorities that do not exist. The solicitor has informed the court the list and summary were prepared using an artificial intelligence (“AI”) tool incorporated in the legal practice management software he subscribes to. The solicitor acknowledges he did not verify the accuracy of the information generated by the research tool before submitting it to the court. 

The solicitor concerned is Mr Dayal, a Victorian solicitor and principal of the firm C Law Firm. I will refer to him as “the solicitor” and his name and the name of his firm will be anonymised when my reasons are published, noting the purpose of my decision is not punitive. 

For the background to the matter, I refer to my earlier ex tempore reasons delivered on 19 July 2024 in the enforcement proceeding in which the solicitor appeared as agent for another firm of solicitors.  Those reasons explain the circumstances in which the list and summary of authorities was tendered by the solicitor, how the content of the list and summary of authorities was identified to be inaccurate and the solicitor’s acknowledgement that the list and summary of authorities was prepared with the assistance of AI.... 

In his written submissions the solicitor acknowledged: (a) Handing up to the court on 19 July 2024 a document that purported to contain summaries of relevant authorities and included what looked like medium neutral citations identifying those decisions; (b) Using legal software, and in particular an AI driven research tool module, to generate the list of authorities and summaries; (c) Neither he nor another legal practitioner had reviewed the output generated by the research tool to ensure the accuracy of the list of authorities and case summaries; and (d) The authorities identified in the list and summary tendered to the court do not exist. 

The solicitor has offered an unconditional apology to the court for tendering the inaccurate list and summary of authorities. He has provided an assurance that he will “take the lessons learned to heart and will not commit any such further breach of professional standards in the future.” He asks that I not make a referral to the Victorian Legal Services Board. 

The submissions made by the solicitor include that he did not intentionally mislead the court. In support of that submission, the solicitor provided information as to the circumstances which lead to him relying on the AI tool within the practice management software he uses and how he generated the list of authorities and case summaries. He explained that he did not fully understand how the research tool worked. He acknowledged the need to verify AI assisted research, or indeed any source of legal research relied upon, for accuracy and integrity. 

The solicitor outlined the steps he has taken to address and mitigate the impact of his conduct, including voluntarily making a payment to the solicitors for the other party in the enforcement proceeding, in settlement of costs thrown away for the hearing on 19 July 2024. He says he has informed the Legal Practitioners Liability Committee (“LPLC”) of what occurred and that the LPLC is providing him with ongoing professional support. The solicitor has also provided submissions in relation to his personal and professional circumstances and the stress and cost caused to him as a result of his conduct on 19 July 2024. He offered to provide an affidavit to verify the information provided in his submissions. 

Use of AI in litigation 

The use of technology is an integral part of efficient modern legal practice. At the frontier of technological advances in legal practice and the conduct of litigation is the use of AI. Whilst the use of AI tools offer opportunities for legal practitioners, it also comes with significant risks. 

Relevantly to this case, the USA District Court case of Mata v Avianca Inc drew worldwide attention to the risk of relying on generative AI for research purposes in litigation without independent verification. In that case, attorneys of a firm who relied on generative AI to prepare legal submissions which were filed referring to non-existent cases, and initially stood by the submissions when called into question by the court, were found to have abandoned their professional responsibilities and sanctioned. The USA District Court outlined the potential harms flowing from the filing of bogus submissions in its judgment as follows:  

Many harms flow from the submission of fake opinions. The opposing party wastes time and money in exposing the deception. The Court’s time is taken from other important endeavors. The client may be deprived of arguments based on authentic judicial precedents. There is potential harm to the reputation of judges and courts whose names are falsely invoked as authors of the bogus opinions and to the reputation of a party attributed with fictional conduct. It promotes cynicism about the legal profession and the American judicial system. And a future litigant may be tempted to defy a judicial ruling by disingenuously claiming doubt about its authenticity.

The potential harms identified by the USA District Court apply to the reliance on non-existent authorities in this court. 

Whilst this court has not yet done so, a number of courts in Australia and overseas have formulated guidelines for the responsible use of generative AI by litigants and lawyers, to assist those conducting litigation before them. 

Guidelines issued by each of the Supreme Cout of Victoria and County Court of Victoria for example,emphasise: (a) Parties and practitioners who are using AI tools in the course of litigation should ensure they have an understanding of the manner in which those tools work, as well as their limitations; (b) The use of AI programs must not indirectly mislead another participant in the litigation process (including the court) as to the nature of any work undertaken or the content produced by that program. Ordinarily parties and their practitioners should disclose to each other the assistance provided by AI programs to the legal task undertaken;  and (c) The use of AI to assist in the completion of legal tasks must be subject to the obligations of legal practitioners in the conduct of litigation, including the obligation of candour to the court. 

Importantly in the context of this matter, the guidelines issued by the Supreme Court and County Court of Victoria explain that generative AI and large language models create output that is not the product of reasoning and nor are they a legal research tool. Generative AI does not relieve the responsible legal practitioner of the need to exercise judgment and professional skill in reviewing the final product to be provided to the court. 

Duties of legal practitioners 

Whilst not issued by this court or applying directly to practitioners conducting litigation in this court, I mention these particular guidelines because they reflect the responsible use of AI by practitioners in litigation by reference to the duties of legal practitioners generally, including the duty not to mislead the court or another participant in the litigation process and the duty of candour to the court. In that sense, the guidance provided by these particular guidelines is applicable to practitioners conducting litigation in this court. 

Relevantly to the conduct of the solicitor before me, the duties of Victorian solicitors include: (a) The paramount duty to the court and to the administration of justice,  which includes a specific duty not to deceive or knowingly or recklessly mislead the court; (b) Other fundamental ethical duties, including to deliver legal services competently and diligently;  and (c) To not engage in conduct which is likely to diminish public confidence in the administration of justice or bring the legal profession into disrepute. 

The solicitor has acknowledged a breach of the professional standards expected of a solicitor in this court, by his conduct in tendering a list and summary of authorities that do not exist, generated without disclosing the source of the information presented to the court and without verifying its accuracy.

Humpreys J stated that 'it is in the public interest for the Victorian Legal Services Board and Commissioner to be aware of the professional conduct issues arising in this matter, given the increasing use of AI tools by legal practitioners in litigation more generally'.

In the earlier judgment - Handa & Mallick [2024] FedCFamC2F 957 - Humphreys J stated 

 The matter was stood down this morning for the purposes of the parties’ legal representatives discussing the issues identified earlier this morning in relation to the enforcement application and to see if there was any prospect of a negotiated resolution. I asked if either party’s representative was in a position to provide me with any authorities that they sought to rely upon, for me to read while the matter was stood down. 

Mr B tendered a single-page list of authorities. Upon returning to chambers neither I nor my associates were able to locate the cases identified in that list. The case citations provided for each of the four listed cases correspond with cases reported by names. My associates asked Mr B to provide copies of the authorities referred to in the list, and he did not do so. 

When the matter returned to court, I asked Mr B if the list of authorities had been provided using artificial intelligence. He informed me the list had been prepared from LEAP, being a legal software package, as I understand it, used for legal practice management and other purposes. I asked if LEAP relies on artificial intelligence. He indicated that it does, answering “there is an artificial intelligence for LEAP.” I foreshadowed making procedural orders later in the day, requiring Mr B to provide an explanation as to what had occurred. Mr B clarified this afternoon that he prepared the list of authorities and not Ms Aus Lawyers. 

I informed the parties and their legal representatives this morning that as a concern had arisen in relation to the veracity of information provided in the list of authorities, a concern had in turn been raised in relation to the competency and ethics of Mr B. In light of what transpired, I asked that the husband be assisted to seek advice from a duty lawyer in relation to Mr B continuing to assist him today. The husband has been present in court throughout these discussions. He informed the court via Mr B and also directly in court after seeing the duty lawyer that he is comfortable for Mr B to continue assisting him today. 

Unfortunately, the parties have been unable to reach an agreement in relation to the enforcement application. That may be because other matters have arisen during the course of the day taking their attention and time away from their negotiations. I encourage them to continue those negotiations over the coming days, pending the adjourned hearing next Wednesday. 

I have foreshadowed with Mr B and counsel for the wife, making an order providing Mr B an opportunity to respond to the court's proposal to refer his conduct in tendering the apparently inaccurate list of authorities today, to the Legal Services Board and Commissioner for investigation. Beyond that, I will not be making an assessment or a determination in relation to that conduct. That will be a matter for the legal professional body if a referral is made. The purpose of the order I make is for Mr B to be afforded procedural fairness in relation to my proposal to make that referral. I will provide him with one month to do that. Mr B has been informed of the orders I intend to make this afternoon and has not wished to make submissions against that course. 

Counsel for the wife has foreshadowed making an application for costs in relation to the adjournment of today's hearing. He anticipates doing so at the conclusion of the enforcement hearing rather than separately today. I have indicated to the parties and to their legal representatives today, that if any application is made for costs to be paid personally by Mr B (as agent appearing today for the husband), he is to be put on notice of that application and have an opportunity to respond by way of procedural fairness. I will ensure that any further orders made in relation to the foreshadowed cost application provide for that to occur.

09 October 2024

AI Liability

'U.S. Tort Liability for Large-Scale Artificial Intelligence Damages: A Primer for Developers and Policymakers' (RAND, 2024) by Ketan Ramakrishnan, Gregory Smith and Conor Downey comments 

Leading artificial intelligence (AI) developers and researchers, as well as government officials and policymakers, are investigating the harms that advanced AI systems might cause. In this report, the authors describe the basic features of U.S. tort law and analyze their significance for the liability of AI developers whose models inflict, or are used to inflict, large-scale harm. 

Highly capable AI systems are a growing presence in widely used consumer products, industrial and military enterprise, and critical societal infrastructure. Such systems may soon become a significant presence in tort cases as well—especially if their ability to engage in autonomous or semi-autonomous behavior, or their potential for harmful misuse, grows over the coming years. 

The authors find that AI developers face considerable liability exposure under U.S. tort law for harms caused by their models, particularly if those models are developed or released without utilizing rigorous safety procedures and industry-leading safety practices. 

At the same time, however, developers can mitigate their exposure by taking rigorous precautions and heightened care in developing, storing, and releasing advanced AI systems. By taking due care, developers can reduce the risk that their activities will cause harm to other people and reduce the risk that they will be held liable if their activities do cause such harm. 

The report is intended to be useful to AI developers, policymakers, and other nonlegal audiences who wish to understand the liability exposure that AI development may entail and how this exposure might be mitigated.

 Key Findings are 

• Tort law is a significant source of legal risk for developers that do not take adequate precautions to guard against causing harm when developing, storing, testing, or deploying advanced AI systems. 

Under existing tort law, there is a general duty to take reasonable care not to cause harm to the person or property of others. This duty applies by default to AI developers—even if targeted liability rules to govern AI development are never enacted by legislatures or regulatory agencies. In Chapter 3, we discuss the requirements of the duty to take reasonable care, and how AI developers might comply (or fail to comply) with these requirements. 

• There is substantial uncertainty, in important respects, about how existing tort doctrine will be applied to AI development. 

Jurisdictional variation and uncertainty about how legal standards will be interpreted and applied may generate substantial liability risk and costly legal battles for AI developers. Courts in different states may reach different conclusions on important issues of tort doctrine, especially in novel fact situations. Tort law varies significantly across both domestic and international jurisdictions. In the United States, each state has a different body of tort law, which coexists alongside federal tort law. Which state’s tort law applies to a dispute depends on complex choice-of-law rules, which in turn depend on the location of the tortious harm at issue (among other things). Moreover, tort decisions often depend on highly context-specific applications of broad legal standards (such as the negligence standard of “reasonable care”) by lay juries. As a result, tort liability can be difficult to predict, particularly with respect to emergent technologies that pose novel legal questions. In the wake of large-scale harms with effects spread across many states, AI developers may face many costly suits across multiple jurisdictions, each with potentially different liability rules. The tort liability incurred by irresponsible AI development may be sufficiently onerous, in the case of sufficiently large-scale damage, to render an AI developer insolvent or force it to declare bankruptcy. Given the cost and risk of litigating a plausible tort suit, moreover, there will often be strong financial incentive for an AI developer (or its liability insurer) to agree to a costly settlement before a verdict is reached. 

• AI developers that do not employ industry-leading safety practices, such as rigorous red- teaming and safety testing or the installation of robust safeguards against misuse, among others,may substantially increase their liability exposure. 

Tort law gives significant credit to industry custom, standards, and practice when determining whether an agent has acted negligently (and is thus liable for the harms it has caused). If most or many industry actors take a certain sort of precaution, this fact will typically be regarded as strong evidence that failing to take this precaution is negligent. Developers who forgo common safety practices in the AI industry, without instituting comparably rigorous safety practices in their stead, may thus increase the likelihood that they will be found negligent should their models cause or enable harm. Therefore, AI developers may wish to consider employing state-of-the-art safety procedures by, for instance, evaluating models for dangerous capabilities, fine-tuning models to limit unsafe behavior, monitoring and moderating models hosted via an application programming interface (API) for dangerous behavior, investing in strong information security measures for model weights, installing reasonably robust safeguards against misuse in potentially dangerous AI systems, and releasing these systems in ways that minimize the chance that third parties will remove the safeguards installed in them. 

• While developers face significant liability exposure from the risk that third parties will misuse their models, there is considerable uncertainty about how this issue will be treated in the courts, and different states may take markedly different approaches. 

Most American courts today maintain that a defendant will be liable for negligently enabling a third party to cause harm, maliciously or inadvertently, if this possibility was reasonably foreseeable to the defendant. But “foreseeability” is a pliable concept, and in practice some courts will only hold a defendant liable for enabling third-party misbehavior if such behavior was readily or especially foreseeable. The risks of misuse of advanced AI systems are being actively discussed and debated, and several leading AI developers take significant precautions to guard against such risks; these facts will tend to support the determination that such misuse was foreseeable, in the event that it occurs. The fact that many of these risks are of a novel kind, and have not previously materialized, may cut in the opposite direction. In some cases, moreover, courts may decline to hold defendants liable for negligently enabling third parties to cause harm even when the possibility of such misuse is foreseeable. For these reasons, and others, there is a good deal of uncertainty about how liability for third-party misuse will be adjudicated in the courts. It would not be surprising if different states took different positions on this issue, just as different states have taken different positions on liability for enabling the misuse of other dangerous instrumentalities (such as guns). Thus, a careless AI developer could face a series of complex and costly legal battles if its model is misused to inflict harm across many jurisdictions. 

• Safety-focused policymakers, developers, and advocates can strengthen AI developers’ incentives to employ cutting-edge safety techniques by developing, implementing, and publicizing new safety procedures and by formally promulgating these standards and procedures through industry bodies. 

The popularization and proliferation of safe and secure AI development practices by safety-conscious developers and industry bodies can help set industry standards and “customs” that courts may consider when evaluating the liability of other developers, creating stronger incentives for safe and responsible AI development. 

• Policymakers may wish to clarify or modify liability standards for AI developers and/or develop complementary regulatory standards for AI development. 

Our analysis suggests that there remains significant uncertainty as to how existing liability law will be applied if harms are caused by advanced AI systems. This uncertainty could conceivably lead some developers to be too cautious, while pushing other developers to neglect the liability risks associated with unsafe development. Clarifying or modifying liability law might thus facilitate responsible innovation and increase the tort system’s ability to incentivize safe behavior. Legislation might also help to remedy the inherent limitations of the tort liability system. For example, tort liability cannot easily address the fact that certain AI developers might discount serious risks on the basis of idiosyncratic views, or that an AI company’s liability exposure— which is limited by its total assets—might fail to provide it with adequately strong incentives for taking due care. Carefully designed legislation might remedy these shortcomings through the creation of a well-tailored regulatory regime, the clarification or improvement of existing liability law to more clearly identify when a developer or another party is liable for harms, or the establishment of minimum safety requirements for forms of AI development that pose especially significant risks to national security or public welfare.

08 October 2024

Vitalism

'African vital force and the permissibility of euthanasia' by Kirk Lougheed in (2024) Journal of Medical Ethics comments 

One prominent idea in African metaphysics says that everything that exists, including both animate and inanimate objects, is imbued with an imperceptible energy known as vital force or vitalism. On this view, the goal of morality is to preserve and strengthen the force in oneself and in others. It is life itself that is the object of the greatest moral concern since once a person dies their force is completely snuffed out, which is considered to be the worst possible outcome. Inasmuch as Africans across the sub-Saharan affirm vital force, the sociological data suggesting that there is little support for euthanasia or physician-assisted suicide on the African continent is unsurprising. Instead, what is surprising is that in the few places where euthanasia is discussed by African ethicists, it often receives a very sympathetic hearing.  Perhaps even more surprising is that the permissibility of euthanasia has recently been defended by appealing to an ethic based on African vital force as located in the work of Molefe and Maraganedzha. They suggest that when vital force cannot be participated in or otherwise fostered, then euthanasia is permissible. This is particularly so when such a person cannot meaningfully participate in the life of the community, which is regarded as a particularly powerful way of enacting vital force. Even for those who do not believe in the existence of vital force, there is still likely a certain amount of intuitive plausibility to the claim that relating well with family and friends is one of the most important activities for conferring meaning to one’s life. 

Before proceeding, it is important to clarify the scope of my discussion, since euthanasia is a broad term that captures many different ideas. First, there is a distinction between passive and active euthanasia, where the former involves withholding life-saving treatment and the latter occurs when a patient’s life is actively terminated. Passive euthanasia occurs frequently and is not what I have in view here. For example, a patient with an advanced form of cancer may decline doing another round of treatment even if doing so would almost certainly extend their life, if only briefly. Active euthanasia is usually the target of the current philosophical debate. This can involve cases where a physician directly injects a patient with a lethal dose of medication or where a lethal dose is prescribed, but the patient must administer it themselves. This is sometimes referred to as Medical Assistance in Dying (or ‘MAiD’).i I am primarily concerned with whether it is permissible for a patient to use euthanasia and so what I say is meant to be consistent with both forms. In other words, my claims are intended to apply to a patient requesting that a doctor inject them with a lethal dose of medication or prescribe them a lethal dose that they can take themselves.ii Finally, there is a distinction among the voluntary, non-voluntary and involuntary active euthanasia. Voluntary active euthanasia occurs when a patient is competent, and requests MAiD.iii Non-voluntary active euthanasia occurs when the patient is incapacitated, and the request comes from a legitimate proxy or advanced directive and is fulfilling the patient’s wishes.iv Finally, involuntary active euthanasia occurs when euthanasia is conducted against the patient’s wishes.v I am primarily concerned with evaluating how well Molefe and Maraganedzha’s argument can defend voluntary active euthanasia, where the patient is competent and makes a request for MAiD. In what follows, when I write of ‘euthanasia’, I mean active voluntary euthanasia unless stated otherwise. 

In view of my focus, notice that I am primarily concerned with the moral question of whether it is permissible for an agent who is considering euthanasia to be euthanised. In other words, is it permissible for the patient themselves to request and receive MAiD? Nothing that I write is intended to suggest that this is the only angle worthy of moral reflection when it comes to euthanasia. Family, close friends and medical practitioners of someone who is seriously ill may be forced to confront questions about euthanising someone else. The set of relevant moral considerations may not be identical to those of the agent who is considering it for themselves. It is not that I think Molefe and Maraganedzha’s argument suggests nothing worthy of discussion from these other perspectives; but this paper would quickly become far too long without delineating the scope in this way. Though I will suggest (see The nature of other-regarding duties in African thought) that the agent considering euthanasia needs to examine their decision from within the context of their community, I am still asking what is permissible for the agent. 

In what follows, after reconstructing the argument (see Reconstructing the argument for euthanasia from vital force), I point to a number of areas of potential concern (see Evaluating the argument for euthanasia based on vital force). While these objections do not show that euthanasia is impermissible, they do suggest that it is doubtful its permissibility can be defended on the basis of African vital force. I conclude (see Conclusion) by observing that the time is ripe for cross-cultural dialogue between the Anglo-American and African bioethical traditions.