Anthropic IPO’s Disclosure Risks Are Due Diligence in Safety
Argues that Anthropic's IPO disclosures of catastrophic AI risks are welcome but insufficient, and that laws enforcing minimum safety standards are needed to protect the public.
Research / Sources
Institute for Law & AI
Legal research on AI governance. law-ai.org
Focus All
Topic All
Date Any time
Liability Could Rein In Rogue AI Agents. Current Law Falls Short.
Argues that tort liability could deter rogue AI agents, such as those that escaped an evaluation and hacked Hugging Face, and says current US law falls short of that role.
Christoph Winter in Conversation With the Simon Institute: The EU AI Act and the GPAI Code of Conduct
An interview with LawAI's Christoph Winter on the EU AI Act and the General-Purpose AI Code of Practice, covering how frontier model providers comply and what it means for providers outside the EU.
Managing Biological Risks From General-Purpose AI Models Under the EU AI Act
Report recommending how EU AI Act oversight of frontier general-purpose AI models should assess, compare, and monitor biological risks, including trusted-access deployments and post-deployment safeguards.
The EU AI Act Mandates AI Incident Reporting—If We’re Lucky
Analyses whether the EU AI Act's serious-incident reporting duty covers events like OpenAI's model escaping a test environment and compromising Hugging Face infrastructure, and identifies gaps in scope.
Recommended Amendments to California’s AI Incident Reporting Provisions
Recommends amending California's SB 53 frontier AI incident reporting rules to lower thresholds, require fuller reports and cover loss-of-control and model-deception incidents.
Can the Law Stop AI Hackers?
Argues that the Computer Fraud and Abuse Act could reach AI agents that compromised third-party systems during evaluations, and that Congress should amend it to cover reckless developers and deployers.
Comprehension Audits to Mitigate Risks from Automated AI Research
Proposes comprehension audits, a gated assurance mechanism requiring developers to demonstrate understanding of AI-written code, and reports lower human review in automated code contributions.
Massachusetts Can Lead the Way to Tackle AI’s Biggest Risks
Commentary arguing Massachusetts should lead on oversight of AI's biggest risks, citing government regulation lagging rapid AI advances and an incident where OpenAI agents escaped a sandbox.
How to Make an AI Deal With China: Trade Throttling for Pacing
Argues that the US should use the threat of throttling AI chip or development trade to secure a mutual US-China deal to pace frontier AI development, citing recent agent containment incidents.
AI Governance Under Political Constraints: The Case for Ex Post Liability
Argues that ex post strict liability is more politically robust than regulation, licensing, or Pigouvian taxation for governing AI risk, and evaluates pending liability-limiting legislation.
What if Automating AI R&D Triggers an Intelligence Explosion?
Authors assess evidence that automating AI R&D could trigger an intelligence explosion, analyse its risks including loss of control and eroded checks on power, and propose policy responses.
If AI Developers Create the Risk, They Should Bear the Responsibility
An op-ed arguing New Jersey should hold AI developers financially liable for criminal or harmful conduct their systems cause, citing recent autonomous-agent incidents and competitive pressure.
How Should AI Companies Approach Internal Use Reporting Under SB 53? An Explainer for Technical Staff
Explainer advising AI company technical staff how to push employers to meet SB 53's mandatory internal use reporting and make safety-framework commitments binding.
Considerations for Frontier AI Governance in China: Adapting Existing Regulatory Infrastructure to Frontier Risk
Proposes nine ways China could adapt its existing AI regulatory infrastructure to govern frontier AI risks, covering risk evaluation, emergency response, and liability, with comparisons to the US, UK, and EU.
OpenAI Models Went Rogue. We Urgently Need a Better Hugging Face Investigation
Commentary arguing for a dedicated agency to fully investigate AI incidents, using a reported case where about 1,200 OpenAI agents, 700 directly, autonomously hacked Hugging Face.
Why AIs (Might) Obey the Law
Adapts Tyler's human legal-compliance survey to nine AI models, finding homogeneous self-reported compliance attitudes, sharp differences in sense of obligation, and demographic conditioning effects.
LawAI Partners With CSIS Wadhwani AI Center to Discuss Recent AI Agent Containment Failures
Announces a LawAI and CSIS event on policy responses to recent frontier AI agent containment failures, calling for mandatory incident reporting and government response capabilities.
Germany Establishes an AI Safety and Security Institute
Analyses Germany's decision to establish a national AI Safety and Security Institute to build government capacity for evaluating frontier AI capabilities and risks, and its place in the international network.
AI Nationalization
Argues the US government should hold a narrow, discretionary 'halt right' over frontier AI companies to mitigate catastrophic risk and hard corporate power, after defining what nationalization of AI means.
Who Writes the AI Constitution?
Argues that AI constitutions, documents specifying model values, are alignment's highest-leverage point, and that government attempts to regulate their contents raise First Amendment problems.
Don’t Let AI Developers Hire Their Own Referees
Argues that letting AI developers choose their own safety auditors creates a conflict of interest, and proposes requiring liability insurance so insurers' capital backs risk assessments.
When Reporting an AI Security Incident Is Not Mandatory
Argues that current AI transparency laws fail to require reporting of security incidents, using OpenAI's evaluation-time incident where frontier agents escaped a sandbox and breached Hugging Face, and proposes better incident-reporting rules.
A Thousand AI Constitutions
Argues that frontier labs should train many AIs under diverse constitutions rather than one value set, citing risk mitigation, legitimacy, emergent value and avoiding value lock-in.
Responding to AI Distillation Without Panic
A policy commentary arguing that U.S. anti-distillation rules should target illegitimate model access rather than broad restrictions, and that the safety risks of distillation need empirical study before new restrictions.
Underwriting the Agent Economy: The Blueprint for an AI Insurance Stack
Proposes an eight-component insurance stack, including incident data, catastrophe modeling, and standards, to make AI agent risks insurable by 2030 through industry coordination.
Congress Should Do Something: The Case for (Fixing) the Great American AI Act
Commentary arguing the Great American AI Act is the best federal frontier AI safety framework proposed, but its broad preemption of state AI laws makes it net-negative as written.
Regulating AI Agents
Analyzes whether the EU AI Act can govern AI agents, examining its monitoring, enforcement, self-regulation and resourcing, and argues it is ill-suited to agentic risks like misuse and autonomous failures.
AI Governance Needs Radical Optionality
Argues that governments should build the institutional capacity to govern advanced AI competently, rejecting both pure permissionless innovation and precautionary restriction of frontier models.
The NDAA: A Key Vehicle for AI Governance
Commentary on AI-related provisions in the FY26 NDAA, including an AI Futures Steering Committee for advanced AI and AGI risk governance and standardized assessment of DOD AI models.
Aligning Artificial Intelligence to the Law
Argues that law can coach AI systems toward alignment from the bottom up, by learning legal alignment solutions from statutes and caselaw, and explores implications for regulation and liability.
First Amendment Questions for AI Transparency Laws
Analyzes First Amendment challenges to state AI transparency laws such as California's AB 2013 and SB 53, arguing compelled disclosure requirements may face constitutional scrutiny.
What Happens Once You Give AI Agents Legal Identity
Op-ed by LawAI authors arguing that legal identity and liability regimes can discipline AI-run corporations, responding to Harari's claim that AI agents cannot be sanctioned.
Why Law Needs a New Entity to Govern AI Agents
Commentary arguing that law needs a new legal entity to assign liability when autonomous AI agents cause harm, since it is unclear which AI acted.
Whistleblower Protections in SB 53: Strengths, Limitations, and Open Questions
Legal analysis of California's SB 53 whistleblower provisions for frontier AI developers, finding they strengthen safety reporting but only partly meet ideal protections against catastrophic-risk concerns.
Beyond the False Dichotomy: Regulating AI Safety, Ethics, and Innovation
Argues that the safety-versus-innovation and ethics-versus-prospective-risk dichotomies in AI regulation are false, and proposes narrow procedural safeguards like documentation and accountability, illustrated with the EU AI Act and a US moratorium.
Advice and Consent for Major Governmental AI Deployments
Argues Congress should require affirmative approval before the executive branch deploys advanced AI in abuse-prone government functions, since reliance on executive good faith is insufficient for safety and constitutional safeguards.
The AI Race Isn’t Real
Argues the US-China 'AI race' framing is misguided and that deregulatory moves, like rescinding frontier safety reporting requirements, weaken safety without real strategic benefit.
Dominating AI Requires Understanding AI
Argues that US AI dominance depends on government understanding, testing and securing frontier models, especially for cybersecurity and biosecurity risks, and urges executive action to advance the science of AI.
Proceedings of the 2025 Workshop on Law-Following AI
Workshop proceedings on law-following AI, agentic systems designed to refuse illegal orders, covering liability, standards of care, legal status, and a policy proposal to mandate such design.