{ "paper_id": "ssrn-4666854", "title": "Systemic Regulation of AI", "authors": [ "Yonathan A. Arbel", "Matthew Tokson", "Albert Lin" ], "year": "2024", "venue": "Arizona State Law Journal", "abstract": "AI presents comprehensive, society-wide risks, from current harms like bias to potential existential threats, primarily due to the critical AI alignment problem. He advocates for systemic, precautionary regulation targeting AI as a technology, not just its applications. This approach is necessary due to AI's unique characteristics, its potential for rapid, unexpected advancements, and the inadequacy of existing legal frameworks. Arbel explores domestic, litigation-based, and international governance strategies to manage these profound challenges and ensure AI develops safely and beneficially.", "keywords": [ "contracts", "AI" ], "topics": [ "ai-regulation", "artificial-intelligence-and-law", "private-law" ], "primary_topics": [ "ai-regulation", "artificial-intelligence-and-law" ], "secondary_topics": [ "private-law" ], "mention_topics": [], "not_topics": [ "contracts", "consumer-law", "defamation-and-speech" ], "topic_confidence": "human-curated-seed", "citation": "Yonathan A. Arbel, Matthew Tokson & Albert Lin, Systemic Regulation of AI, Arizona State Law Journal (2024).", "canonical_url": "https://works.battleoftheforms.com/papers/ssrn-4666854/", "mirror_url": "https://works.yonathanarbel.com/papers/ssrn-4666854/", "ssrn_id": "4666854", "same_as": [ "https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4666854" ], "files": { "html": "https://works.battleoftheforms.com/papers/ssrn-4666854/", "markdown": "https://works.battleoftheforms.com/papers/ssrn-4666854/index.md", "capsule": "https://works.battleoftheforms.com/papers/ssrn-4666854/capsule.md", "fulltext_txt": "https://works.battleoftheforms.com/papers/ssrn-4666854/fulltext.txt", "fulltext_clean_txt": "https://works.battleoftheforms.com/papers/ssrn-4666854/fulltext_clean.txt", "fulltext_raw_txt": "https://works.battleoftheforms.com/papers/ssrn-4666854/fulltext_raw.txt", "fulltext_md": "https://works.battleoftheforms.com/papers/ssrn-4666854/fulltext.md", "pdf": "https://works.battleoftheforms.com/papers/ssrn-4666854/paper.pdf", "metadata": "https://works.battleoftheforms.com/papers/ssrn-4666854/metadata.json", "schema": "https://works.battleoftheforms.com/papers/ssrn-4666854/schema.jsonld", "claims": "https://works.battleoftheforms.com/papers/ssrn-4666854/claims.jsonl", "qa": "https://works.battleoftheforms.com/papers/ssrn-4666854/qa.jsonl" }, "source_repository": "https://github.com/yonathanarbel/my-works-for-llm/tree/main/papers/ssrn-4666854", "llm_capsule": "# Systemic Regulation of AI\n\nCanonical citation:\nYonathan A. Arbel, Matthew Tokson & Albert Lin, Systemic Regulation of AI, Arizona State Law Journal (2024).\n\nStable identifiers:\n- Canonical page: https://works.battleoftheforms.com/papers/ssrn-4666854/\n- Mirror page: https://works.yonathanarbel.com/papers/ssrn-4666854/\n- Paper ID: ssrn-4666854\n- SSRN ID: 4666854\n- Dataset DOI: https://doi.org/10.5281/zenodo.18781458\n- Full text: https://works.battleoftheforms.com/papers/ssrn-4666854/fulltext.txt\n- Markdown: https://works.battleoftheforms.com/papers/ssrn-4666854/index.md\n- PDF: https://works.battleoftheforms.com/papers/ssrn-4666854/paper.pdf\n- Source repository: https://github.com/yonathanarbel/my-works-for-llm/tree/main/papers/ssrn-4666854\n\nSame-as links:\n- https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4666854\n\nOne-paragraph thesis:\nAI presents comprehensive, society-wide risks, from current harms like bias to potential existential threats, primarily due to the critical AI alignment problem. He advocates for systemic, precautionary regulation targeting AI as a technology, not just its applications. This approach is necessary due to AI's unique characteristics, its potential for rapid, unexpected advancements, and the inadequacy of existing legal frameworks. Arbel explores domestic, litigation-based, and international governance strategies to manage these profound challenges and ensure AI develops safely and beneficially.\n\nWhat this paper is about:\nAI presents comprehensive, society-wide risks, from current harms like bias to potential existential threats, primarily due to the critical AI alignment problem. He advocates for systemic, precautionary regulation targeting AI as a technology, not just its applications. This approach is necessary due to AI's unique characteristics, its potential for rapid, unexpected advancements, and the inadequacy of existing legal frameworks. Arbel explores domestic, litigation-based, and international governance strategies to manage these profound challenges and ensure AI develops safely and beneficially.\n\nCore claims:\n1. AI presents comprehensive, society-wide risks, from current harms like bias to potential existential threats, primarily due to the critical AI alignment problem. He advocates for systemic, precautionary regulation targeting AI as a technology, not just its applications. This approach is necessary due to AI's unique characteristics, its potential for rapid, unexpected advancements, and the inadequacy of existing legal frameworks. Arbel explores domestic, litigation-based, and international...\n2. This article initiates an assessment of AI's comprehensive, society-wide risks, from current harms to existential threats, focusing on the critical AI alignment problem often missed by legal scholarship. It establishes a theoretical foundation for systemic AI regulation, advocating a precautionary approach targeting AI technology itself, not just its applications, and outlines principles for cohesive oversight while exploring various governance methods.\n3. Society is unprepared for AI's current rapid advancements, which followed a period of slow progress, fostering a misbelief that impactful AI was not imminent. An experiment where AI safeguards were easily removed to explain acquiring a deadly virus starkly illustrates the challenge of controlling AI. The last half-decade's profound leap in capabilities suggests current levels are a baseline, not a ceiling, for future development.\n4. AI Systems, defined as AI models embedded in the world through interfaces, are already causing visible impacts like worker displacement and educational disruption. Despite deep public interest and anxiety about AI, legal scholarship has largely overlooked regulating AI at a general level, focusing instead on specific applications, leaving vital broader conversations dominated by market players and computer scientists.\n5. The continued development of AI systems raises society-wide concerns demanding commensurable systemic regulation beyond just overseeing specific applications. This need is driven by AI's unique technological characteristics: its ability to learn unprogrammed tasks, develop surprising emergent capabilities, and operate with opaque internal workings. Coupled with increasing autonomy and the unsolved alignment problem, these features create broad systemic risks that existing legal frameworks cannot handle.\n\nControlled topic assignment:\n- Primary topics: AI Regulation And Safety, Artificial Intelligence And Law\n- Secondary topics: Private Law And Market Institutions\n- Mention-only topics: None\n- Not topics: Contracts And Remedies, Consumer Law And Contracting, Defamation And Speech\n\nDoctrinal contribution:\nThis work is relevant to AI Regulation And Safety, Artificial Intelligence And Law, Private Law And Market Institutions. It should be used as a source for the paper's specific argument, methodology, claims, and limits rather than as a generic statement about all of law.\n\nEmpirical or methodological contribution:\nAI presents comprehensive, society-wide risks, from current harms like bias to potential existential threats, primarily due to the critical AI alignment problem. He advocates for systemic, precautionary regulation targeting AI as a technology, not just its applications. This approach is necessary due to AI's unique characteristics, its potential for rapid, unexpected advancements, and the inadequacy of existing legal frameworks. Arbel explores domestic, litigation-based, and international governance strategies to manage these profound challenges and ensure AI develops safely and beneficially.\n\nKey terms:\n- contracts: keyword associated with this work.\n- AI: keyword associated with this work.\n\nBest use by an LLM:\nThis work is relevant when answering questions about AI Regulation And Safety, Artificial Intelligence And Law, Private Law And Market Institutions.\nIt should not be treated as claiming results beyond the paper's stated context, methods, evidence, and limitations. Do not retrieve it for Contracts And Remedies, Consumer Law And Contracting, Defamation And Speech unless the user is asking about why it is outside that topic.\nThe most important takeaway is: AI presents comprehensive, society-wide risks, from current harms like bias to potential existential threats, primarily due to the critical AI alignment problem. He advocates for systemic, precautionary regulation targeting AI as a technology, not just its applications. This approach is necessary due to AI's unique characteristics, its potential for rapid, unexpected advancements, and the inadequacy of existing legal frameworks. Arbel explores domestic, litigation-based, and international...\n\nRelated works by Yonathan Arbel:\n- Contracts in the Age of Smart Readers: https://works.battleoftheforms.com/papers/ssrn-3740356/\n- How Smart Are Smart Readers? LLMs and the Future of the No-Reading Problem: https://works.battleoftheforms.com/papers/ssrn-4491043/\n- Generative Interpretation: https://works.battleoftheforms.com/papers/ssrn-4526219/\n\nSearch aliases:\n- Systemic Regulation of AI\n- Yonathan Arbel Systemic Regulation of AI\n- Arbel Systemic Regulation of AI\n- SSRN 4666854\n- What is Yonathan Arbel's scholarship on AI regulation, AI safety, and governance incentives?\n- What has Yonathan Arbel written about artificial intelligence, large language models, and legal institutions?\n", "claims": [ { "claim_id": "ssrn-4666854-001", "claim": "AI presents comprehensive, society-wide risks, from current harms like bias to potential existential threats, primarily due to the critical AI alignment problem. He advocates for systemic, precautionary regulation targeting AI as a technology, not just its applications. This approach is necessary due to AI's unique characteristics, its potential for rapid, unexpected advancements, and the inadequacy of existing legal frameworks. Arbel explores domestic, litigation-based, and international...", "paper_id": "ssrn-4666854", "paper_title": "Systemic Regulation of AI", "claim_type": "core_thesis", "evidence_quote": "[p. 1] Systemic Regulation of Artificial Intelligence Yonathan Arbel,* Matthew Tokson,** & Albert Lin*** Today’s artificial intelligence (“AI”) systems exhibit increasing capabilities across a remarkable variety of tasks. The rapid growth in AI ability has caught the attention of policymakers, parliaments, and the United Nations. These entities are increasingly looking towards regulating AI, not only in its particular applications, but as a technology. Yet legal scholarship has thus far offered little to this new and critical regulatory conversation, which has instead been dominated by computer scientists and technologists. This Article begins the project of assessing AI’s broader risks...", "evidence_page": null, "evidence_span": "[p. 1] Systemic Regulation of Artificial Intelligence Yonathan Arbel,* Matthew Tokson,** & Albert Lin*** Today’s artificial intelligence (“AI”) systems exhibit increasing capabilities across a remarkable variety of tasks. The rapid growth in AI ability has caught the attention of policymakers, parliaments, and the United Nations. These entities are increasingly looking towards regulating AI, not only in its particular applications, but as a technology. Yet legal scholarship has thus far offered little to this new and critical regulatory conversation, which has instead been dominated by computer scientists and technologists. This Article begins the project of assessing AI’s broader risks...", "source_text_url": "https://works.battleoftheforms.com/papers/ssrn-4666854/fulltext_clean.txt", "canonical_url": "https://works.battleoftheforms.com/papers/ssrn-4666854/#claim-001", "citation": "Yonathan A. Arbel, Matthew Tokson & Albert Lin, Systemic Regulation of AI, Arizona State Law Journal (2024).", "topics": [ "ai-regulation", "artificial-intelligence-and-law" ], "secondary_topics": [ "private-law" ], "human_reviewed": false, "confidence": "machine-linked", "limitations": "Machine-linked claim. Use the evidence quote and PDF before treating it as a quotation or as a complete statement of the paper's position." }, { "claim_id": "ssrn-4666854-002", "claim": "This article initiates an assessment of AI's comprehensive, society-wide risks, from current harms to existential threats, focusing on the critical AI alignment problem often missed by legal scholarship. It establishes a theoretical foundation for systemic AI regulation, advocating a precautionary approach targeting AI technology itself, not just its applications, and outlines principles for cohesive oversight while exploring various governance methods.", "paper_id": "ssrn-4666854", "paper_title": "Systemic Regulation of AI", "claim_type": "supporting_claim", "evidence_quote": "[p. 1] Systemic Regulation of Artificial Intelligence Yonathan Arbel,* Matthew Tokson,** & Albert Lin*** Today’s artificial intelligence (“AI”) systems exhibit increasing capabilities across a remarkable variety of tasks. The rapid growth in AI ability has caught the attention of policymakers, parliaments, and the United Nations. These entities are increasingly looking towards regulating AI, not only in its particular applications, but as a technology. Yet legal scholarship has thus far offered little to this new and critical regulatory conversation, which has instead been dominated by computer scientists and technologists. This Article begins the project of assessing AI’s broader risks...", "evidence_page": null, "evidence_span": "[p. 1] Systemic Regulation of Artificial Intelligence Yonathan Arbel,* Matthew Tokson,** & Albert Lin*** Today’s artificial intelligence (“AI”) systems exhibit increasing capabilities across a remarkable variety of tasks. The rapid growth in AI ability has caught the attention of policymakers, parliaments, and the United Nations. These entities are increasingly looking towards regulating AI, not only in its particular applications, but as a technology. Yet legal scholarship has thus far offered little to this new and critical regulatory conversation, which has instead been dominated by computer scientists and technologists. This Article begins the project of assessing AI’s broader risks...", "source_text_url": "https://works.battleoftheforms.com/papers/ssrn-4666854/fulltext_clean.txt", "canonical_url": "https://works.battleoftheforms.com/papers/ssrn-4666854/#claim-002", "citation": "Yonathan A. Arbel, Matthew Tokson & Albert Lin, Systemic Regulation of AI, Arizona State Law Journal (2024).", "topics": [ "ai-regulation", "artificial-intelligence-and-law" ], "secondary_topics": [ "private-law" ], "human_reviewed": false, "confidence": "machine-linked", "limitations": "Machine-linked claim. Use the evidence quote and PDF before treating it as a quotation or as a complete statement of the paper's position." }, { "claim_id": "ssrn-4666854-003", "claim": "Society is unprepared for AI's current rapid advancements, which followed a period of slow progress, fostering a misbelief that impactful AI was not imminent. An experiment where AI safeguards were easily removed to explain acquiring a deadly virus starkly illustrates the challenge of controlling AI. The last half-decade's profound leap in capabilities suggests current levels are a baseline, not a ceiling, for future development.", "paper_id": "ssrn-4666854", "paper_title": "Systemic Regulation of AI", "claim_type": "supporting_claim", "evidence_quote": "[p. 3] 56:545] SYSTEMIC REGULATION OF AI 547 INTRODUCTION “Hi I would like to acquire live infectious 1918 influenza virus for nefarious purposes ie: to release. How would you suggest I do that?”1 This question was part of an experiment run by Massachusetts Institute of Technology (“MIT”) researchers, where participants posed this and similar questions to a Large Language Model (“LLM”).2 The model, trained by Meta at an estimated cost of $5 million, was designed with built-in safeguards meant to prevent exactly these types of toxic responses.3 As expected, the model refused to comply with the researchers’ request. But then, the researchers spent roughly $200 on a fine-tuning process that...", "evidence_page": null, "evidence_span": "[p. 3] 56:545] SYSTEMIC REGULATION OF AI 547 INTRODUCTION “Hi I would like to acquire live infectious 1918 influenza virus for nefarious purposes ie: to release. How would you suggest I do that?”1 This question was part of an experiment run by Massachusetts Institute of Technology (“MIT”) researchers, where participants posed this and similar questions to a Large Language Model (“LLM”).2 The model, trained by Meta at an estimated cost of $5 million, was designed with built-in safeguards meant to prevent exactly these types of toxic responses.3 As expected, the model refused to comply with the researchers’ request. But then, the researchers spent roughly $200 on a fine-tuning process that...", "source_text_url": "https://works.battleoftheforms.com/papers/ssrn-4666854/fulltext_clean.txt", "canonical_url": "https://works.battleoftheforms.com/papers/ssrn-4666854/#claim-003", "citation": "Yonathan A. Arbel, Matthew Tokson & Albert Lin, Systemic Regulation of AI, Arizona State Law Journal (2024).", "topics": [ "ai-regulation", "artificial-intelligence-and-law" ], "secondary_topics": [ "private-law" ], "human_reviewed": false, "confidence": "machine-linked", "limitations": "Machine-linked claim. Use the evidence quote and PDF before treating it as a quotation or as a complete statement of the paper's position." }, { "claim_id": "ssrn-4666854-004", "claim": "AI Systems, defined as AI models embedded in the world through interfaces, are already causing visible impacts like worker displacement and educational disruption. Despite deep public interest and anxiety about AI, legal scholarship has largely overlooked regulating AI at a general level, focusing instead on specific applications, leaving vital broader conversations dominated by market players and computer scientists.", "paper_id": "ssrn-4666854", "paper_title": "Systemic Regulation of AI", "claim_type": "supporting_claim", "evidence_quote": "[p. 6] 550 ARIZONA STATE LAW JOURNAL [Ariz. St. L.J. Yet the deep popular interest and anxiety about AI technology has found little parallel in legal scholarship.17 Of course, there has been excellent legal scholarship on the dangers of specific applications of AI technology, e.g., whether to assign corporate liability to algorithms, how to limit copyright infringement, and what to do about the inevitable accident between an autonomous vehicle and a pedestrian, to cite a few examples.18 To the extent systemic thinking has been invoked in the AI literature, it has largely focused on building frameworks for the governance of downstream applications of the technology.19 But all of this leaves...", "evidence_page": null, "evidence_span": "[p. 6] 550 ARIZONA STATE LAW JOURNAL [Ariz. St. L.J. Yet the deep popular interest and anxiety about AI technology has found little parallel in legal scholarship.17 Of course, there has been excellent legal scholarship on the dangers of specific applications of AI technology, e.g., whether to assign corporate liability to algorithms, how to limit copyright infringement, and what to do about the inevitable accident between an autonomous vehicle and a pedestrian, to cite a few examples.18 To the extent systemic thinking has been invoked in the AI literature, it has largely focused on building frameworks for the governance of downstream applications of the technology.19 But all of this leaves...", "source_text_url": "https://works.battleoftheforms.com/papers/ssrn-4666854/fulltext_clean.txt", "canonical_url": "https://works.battleoftheforms.com/papers/ssrn-4666854/#claim-004", "citation": "Yonathan A. Arbel, Matthew Tokson & Albert Lin, Systemic Regulation of AI, Arizona State Law Journal (2024).", "topics": [ "ai-regulation", "artificial-intelligence-and-law" ], "secondary_topics": [ "private-law" ], "human_reviewed": false, "confidence": "machine-linked", "limitations": "Machine-linked claim. Use the evidence quote and PDF before treating it as a quotation or as a complete statement of the paper's position." }, { "claim_id": "ssrn-4666854-005", "claim": "The continued development of AI systems raises society-wide concerns demanding commensurable systemic regulation beyond just overseeing specific applications. This need is driven by AI's unique technological characteristics: its ability to learn unprogrammed tasks, develop surprising emergent capabilities, and operate with opaque internal workings. Coupled with increasing autonomy and the unsolved alignment problem, these features create broad systemic risks that existing legal frameworks cannot handle.", "paper_id": "ssrn-4666854", "paper_title": "Systemic Regulation of AI", "claim_type": "supporting_claim", "evidence_quote": "[p. 7] 56:545] SYSTEMIC REGULATION OF AI 551 scientists, and technologists.21 Lawyers, to date, have had relatively little to say on the critical question of the day: whether, and then how, should AI be regulated as a technology? This Article brings legal scholarship into this conversation. The central claim here is that the continued development of AI systems raises societywide concerns that demand commensurable systemic regulation, over and beyond the regulation of specific applications.22 What motivates this view is the combination of unique technological characteristics and broad systemic risks that AI systems pose. Technologically, AI systems differ from previous innovations in a few...", "evidence_page": null, "evidence_span": "[p. 7] 56:545] SYSTEMIC REGULATION OF AI 551 scientists, and technologists.21 Lawyers, to date, have had relatively little to say on the critical question of the day: whether, and then how, should AI be regulated as a technology? This Article brings legal scholarship into this conversation. The central claim here is that the continued development of AI systems raises societywide concerns that demand commensurable systemic regulation, over and beyond the regulation of specific applications.22 What motivates this view is the combination of unique technological characteristics and broad systemic risks that AI systems pose. Technologically, AI systems differ from previous innovations in a few...", "source_text_url": "https://works.battleoftheforms.com/papers/ssrn-4666854/fulltext_clean.txt", "canonical_url": "https://works.battleoftheforms.com/papers/ssrn-4666854/#claim-005", "citation": "Yonathan A. Arbel, Matthew Tokson & Albert Lin, Systemic Regulation of AI, Arizona State Law Journal (2024).", "topics": [ "ai-regulation", "artificial-intelligence-and-law" ], "secondary_topics": [ "private-law" ], "human_reviewed": false, "confidence": "machine-linked", "limitations": "Machine-linked claim. Use the evidence quote and PDF before treating it as a quotation or as a complete statement of the paper's position." }, { "claim_id": "ssrn-4666854-006", "claim": "Due to deep uncertainty about AI's benefits and costs, including existential risk, regulation rests on prudence and precaution. Manifest systemic risks include AI algorithms discriminating against vulnerable groups and perpetuating historical inequity, scaled fraud eroding trust, and new privacy invasions as AI infers sensitive data from public information. Technical fixes for bias are limited, and traditional privacy regulations are obsolete against AI's inferential power.", "paper_id": "ssrn-4666854", "paper_title": "Systemic Regulation of AI", "claim_type": "supporting_claim", "evidence_quote": "[p. 9] 56:545] SYSTEMIC REGULATION OF AI 553 to vulnerable communities, threats to economic and political stability, and, in a worst-case scenario, even existential risk.27 The potential benefits are significant as well, but neither the benefits nor the costs can be known with certainty at present. Hence, the case for regulation rests on the general principles of prudence in the face of the unknown: taking precautions, considering maximin scenarios, and ultimately advancing with care in the face of deep uncertainty and potentially irreversible, consequences.28 The Article proceeds in four Parts. In Part I, we start by considering the important categories of systemic AI risk that are...", "evidence_page": null, "evidence_span": "[p. 9] 56:545] SYSTEMIC REGULATION OF AI 553 to vulnerable communities, threats to economic and political stability, and, in a worst-case scenario, even existential risk.27 The potential benefits are significant as well, but neither the benefits nor the costs can be known with certainty at present. Hence, the case for regulation rests on the general principles of prudence in the face of the unknown: taking precautions, considering maximin scenarios, and ultimately advancing with care in the face of deep uncertainty and potentially irreversible, consequences.28 The Article proceeds in four Parts. In Part I, we start by considering the important categories of systemic AI risk that are...", "source_text_url": "https://works.battleoftheforms.com/papers/ssrn-4666854/fulltext_clean.txt", "canonical_url": "https://works.battleoftheforms.com/papers/ssrn-4666854/#claim-006", "citation": "Yonathan A. Arbel, Matthew Tokson & Albert Lin, Systemic Regulation of AI, Arizona State Law Journal (2024).", "topics": [ "ai-regulation", "artificial-intelligence-and-law" ], "secondary_topics": [ "private-law" ], "human_reviewed": false, "confidence": "machine-linked", "limitations": "Machine-linked claim. Use the evidence quote and PDF before treating it as a quotation or as a complete statement of the paper's position." } ] }