Institute for Law & AI
AI-generated text. This page was generated using artificial intelligence.
The Institute for Law & AI (LawAI) researches and advises on legal responses to artificial intelligence. Its work connects proposals for governing increasingly capable systems with questions about enforceability, legal responsibility and institutional design. The institute describes its principal research teams as US Law and Policy, EU Law, and Legal Frontiers.[1]
Organization and research approach
LawAI traces its origins to the Legal Priorities Project, begun at Harvard Law School in 2019 by Christoph Winter and Eric Martínez with assistance from Cullen O’Keefe and Jonas Freund. It subsequently concentrated on AI. Its institutional account identifies Washington, DC, and Cambridge, UK, as its principal hubs and describes consulting, published research and fellowships as complementary parts of its work.[1]
The institute's EU program addresses general-purpose AI models and enforcement of the EU AI Act and related Code of Practice. Its Legal Frontiers work includes proposals concerning law-following AI. These are research and advisory activities, not legislative or judicial decisions.[1]
Liability as a governance instrument
In The Case for AI Liability, Gabriel Weil argues that liability can create incentives for developers to account for harms imposed on people outside a transaction. His analysis emphasizes third parties who cannot protect themselves by choosing a different supplier. It presents liability as an adaptive mechanism: developers would have incentives to discover precautions rather than simply satisfy a fixed checklist of regulatory requirements.[2]
Weil distinguishes negligence from strict liability. He argues that an inquiry into reasonable precautions can miss upstream choices about whether a system should be developed or deployed at all. His preferred approach would make those responsible for harmful activities bear more of their costs. He also discusses the relationship between state initiatives and federal coordination. These are the author's arguments for reform, not a description of a uniform liability regime already applicable to AI.[2]
Limits and competing analyses
Weil's earlier The Limits of Liability identifies problems that compensation incentives do not adequately solve. Safety research can benefit many firms beyond the organization paying for it, creating a case for public subsidies or prizes. Structural harms and remote causal chains may also be difficult to address through individual claims. For catastrophic losses, a judgment may exceed realistically collectible assets; deterrence based on smaller warning incidents would not work if those warnings never occur.[3]
The same essay considers cross-border enforcement and government conduct. Moving development between jurisdictions can complicate collection of judgments, while officials may respond more to political incentives than to monetary liability. Weil therefore treats liability as part of a broader institutional response rather than a complete answer to every AI-governance problem.[3]
LawAI has also published a contrasting analysis by Daniel Schwarcz and Josephine Wolff. Their October 2025 essay questions whether insurers can reliably price and reduce many generative and agentic AI risks. It highlights sparse incident data, difficulty assessing safeguards, and the possibility that insurers will respond with exclusions or coverage limits instead of better prevention. The publication expressly states that the opinions are the authors' and do not represent the institute's views.[4]
This debate distinguishes compensating victims from preventing harm. Schwarcz and Wolff argue that increasing liability does not automatically supply the data, expertise or incentives required for insurers to supervise complex systems. Their comparison with cyber insurance is a critique of a proposed governance mechanism, not evidence that every category of AI risk is uninsurable.[4]
Selected contributions
- The Limits of Liability (August 2024): public goods, catastrophic loss and enforcement constraints.[3]
- The Case for AI Liability (June 2025): incentives, third-party harms and proposed legal reform.[2]
- The Limits of Regulating AI Safety Through Liability and Insurance (October 2025): a separately attributed critique of insurance-based governance.[4]
Related articles
- United States policy on catastrophic AI risk
- European Union policy on catastrophic AI risk
- EU AI Act
References
- ↑ 1.0 1.1 1.2 1.3 1.4 1.5 Institute for Law & AI, About the Institute. Accessed September 12, 2026.
- ↑ 2.0 2.1 2.2 Gabriel Weil, The Case for AI Liability, June 2025. Accessed September 12, 2026.
- ↑ 3.0 3.1 3.2 Gabriel Weil, The Limits of Liability, August 2024. Accessed September 12, 2026.
- ↑ 4.0 4.1 4.2 Daniel Schwarcz and Josephine Wolff, The Limits of Regulating AI Safety Through Liability and Insurance, October 2025. Accessed September 12, 2026.