links

A growing collection of resources on AI and law for lawyers, policymakers and AI researchers. I try to prioritise primary sources, useful trackers/databases, safety research, and anything that might be useful in bridging the gap between law and technical AI safety.

Start here

AI safety generally asks: “what can go wrong, and how do we reduce the risk?” Law generally asks: “who’s responsible, what duty do they owe, what counts as breach, what evidence proves it, who can enforce it, and what are the consequences?”

I’m a lawyer trying to understand AI

How LLMs work → capabilities → risks.

3blue1brown - neural networks - explainer of one underlying building block of AI
The Physical AI Stack - explainer of the physical aspects of AI
Hertie AI Policy Initiative - explainers on AI policy and governance
MIT AI Risk Initiative - a map of AI risk
Stanford HAI Index - annual report about capabilities, adoption, investment, policy, & social impacts

I work in AI and want to understand law relating to AI

AI legal issues → AI regulation

Institute for Law & AI - works on legal issues related to advanced AI
EU AI Act explorer - overview of EU AI legal framework
NIST AI Risk Management Framework - voluntary US framework for managing AI risks

I’m researching AI governance or safety

Hardware → risk frameworks → regulation → official bodies and third party auditors → incident response.

The Physical AI Stack
MIT AI Risk Initiative - a map of AI risk
EU AI Act explorer - overview of EU AI legal framework
UK - AI Security Institute (AISI) - useful to understand evaluations and model safety
Model Evaluation and Threat Research (METR) - independent nonprofit evaluating frontier AI models' dangerous capabilities

Foundations

Explainers and courses

Andrew Ng - Coursera
3blue1brown - neural networks
Understanding transformers
Sebastian Raschka’s gallery of LLM architectures
BlueDot Impact - AI Governance - free cohort-based course on AI policy and governance
BlueDot Impact - Technical AI Safety - free cohort-based course on technical AI safety
ML4Good - free intensive AI safety bootcamps, with technical and governance tracks
CAIDP - AI Policy Clinic - semester-long, team-based training for future AI policy professionals

Capabilities and trends

AI risks

MIT AI Risk Initiative - 1,700 AI risks
The Human Line Project - AI-induced psychological harm

Law and Regulation

International

International: Council of Europe Framework Convention on Artificial Intelligence The first international treaty on AI. It sets human rights, rule of law, and democratic governance standards (signed by EU, UK, US, and others).

EU

European Union: EU AI Act (and see FLI’s EU AI Act explorer) Tiered risk approach. Bans unacceptable practices (e.g. social scoring, behavioural manipulation), places strict mandates on high-risk AI deployments, and governs general-purpose AI (GPAI) models. Note: the Digital Omnibus moved Annex III high-risk compliance to 2 Dec 2027, and Annex I to 2 Aug 2028.

US

US Federal Executive Order 14365 – Ensuring a National Policy Framework for AI Directs DOJ to challenge state AI laws. No comprehensive federal AI law currently exists.
US State California SB 53 – Transparency in Frontier AI Act Mandates safety frameworks, critical incident reporting, and whistleblower protection. New York RAISE Act Frontier model safety and transparency, closely aligned with California’s SB 53. Effective January 2027. Colorado SB 26-189 – Automated Decision-Making Technology Act Transparency and consumer rights for automated consequential decisions. Texas HB 149 – Texas Responsible AI Governance Act (TRAIGA) Bans certain harmful AI uses and requires government agencies to disclose AI use.

UK, Australia and Asia

South Korea: AI Basic Act Comprehensive AI framework with obligations for high-impact and generative AI.
Vietnam: Law on AI No. 134/2025/QH15 Tiered risk approach. Modelled partly on the EU AI Act. Mandates evaluations for high-impact models, synthetic content watermarking, and incident reporting.

No current dedicated AI law:

Australia: National AI Plan Australia has adopted a distributed regulatory approach in which existing laws (including privacy, consumer protection, online safety, workplace and anti-discrimination law) apply to AI. The government abandoned the 2024 proposal for mandatory guardrails for AI in high-risk settings (including testing, accountability, transparency and human oversight). This does not mean AI is unregulated, enforcement actions have been taken against providers of generative AI.
United Kingdom: No comprehensive AI law, but AI relevant provisions were included in the Data (Use and Access) Act 2025 which introduced looser rules on automated decision-making, provisions on copyright and AI development, and an offence for creating intimate images (deepfakes). An earlier proposed requirement for developers to disclose copyrighted training data was dropped.
China: No comprehensive AI law but the State Council’s 2026 legislative plan calls for comprehensive AI legislation and the Ministry of Justice has said it is working with relevant departments to formulate one. Algorithmic Recommendation Provisions (translation) Regulates internet recommendation and other algorithmic services. Requires transparency and user controls, restricts manipulative and discriminatory uses, and establishes security assessment and algorithm-filing requirements for certain services. Deep Synthesis Provisions (translation) Deepfake rules requiring labelling and consent before editing someone’s biometric features. Interim Measures for Generative AI Services (translation) Regulates generative AI services offered to the public in China. Covers training data, personal information, intellectual property, content moderation and user protection. Services with public-opinion attributes or social-mobilisation capacity must undergo security assessment and algorithm filing. R&D and applications by businesses, universities, research institutions and others that are not providing services to the domestic public are expressly outside scope. Measures for Labeling AI-Generated Synthetic Content (translation) Requires explicit and/or embedded labels for AI-generated or synthetic text, images, audio, video and virtual scenes. Amended Cybersecurity Law (summary) First AI article in a national statute. Introduces an express AI provision into national legislation, requiring the state to support AI research and infrastructure while strengthening AI ethics, risk monitoring, assessment and security supervision.

Trackers

Governance and Safety

Government bodies

AU - AI Safety Institute
AU - National AI Centre
European AI Office
UK - AI Security Institute (AISI)
UK - Incubator for AI
UK - SovereignAI
US - Center for AI Standards & Innovation (CAISI)
US - National Institute of Standards & Technology - see: risk management framework

Organisations

Aden - former diplomats linking AI experts, governments
AU - Centre for AI & Digital Ethics - University of Melbourne - interdisciplinary research on AI and digital ethics
AU - Centre for AI, trust & governance - University of Sydney - socio-technical research on AI trust, governance
AU - Good Ancestors - Australian policy charity on catastrophic risks
Ada Lovelace Institute - independent UK research on AI, society
Anthropic Research - Anthropic’s alignment science research blog
Apollo Research - evaluates AI models for deceptive behaviour
Cambridge University’s Programme on AI Science & Policy - University of Cambridge’s cross-disciplinary AI initiative
Center for AI Safety (CAIS) - catastrophic AI risk research and field-building
Center on Long-Term Risk - research on AI conflict and suffering risks
Center for Law & AI Risk (CLAIR) - legal scholarship on catastrophic AI risk
China - Concordia AI - Beijing-based AI safety and governance consultancy
DeepMind Safety Research - Google DeepMind safety team’s research blog
Forethought - research on navigating the transition to AGI
Frontier Model Forum - industry body of frontier AI developers
GovAI - AI governance research and talent development
Harvard Computational Policy Lab - data science for fairer public policy
Hertie AI Policy Initiative - accessible introductions to AI policy, safety
Institute for AI Policy & Strategy - AI policy research, US national security focus
Institute for Law & AI - legal research on advanced AI governance
Law for AI safety - lawyers using legal tools against AI risks
LawZero - Yoshua Bengio’s non-profit building safe-by-design AI
Machine Intelligence Research Institute (MIRI) - early alignment research institute, now advocacy-focused
Oxford Martin AIGI - Oxford research on global AI governance
Oxford’s AI Readiness in Government - annual Government AI Readiness Index
Princeton Language+Law, Artificial Intelligence, & Society (POLARIS) Lab - Princeton lab on AI, law, society
Palisade Research - demonstrates dangerous AI capabilities to policymakers
Redwood Research - AI control and alignment research
Stanford Regulation, Evaluation and Governance Lab - AI and data science for government
Wharton Generative AI Labs - research on how people use generative AI

Evaluations and standards

AI Verification & Evaluation Research Institute (AVERI)
Guidelight AI Standards
Model Evaluation and Threat Research (METR)
ISO/IEC 42001 - international standard for AI management systems
NIST AI Risk Management Framework - voluntary US framework for managing AI risks

Frontier lab safety policies

AI and Legal Practice

Research

Incident and hallucination databases

AI Hallucination Cases Database - tracking LLM hallucinations in legal cases globally
AI Hallucination Cases: Australia (OBITER)
AI Incident Database - records harms and near harms
Law Library Victoria - hallucinations in case law
OECD AI Incidents Monitor

Legal benchmarks

Legal AI Benchmarking
Legal Hallucination Detection Benchmark
LexBench - Canadian case law
Massive Legal Embedding Benchmark - Isaacus
Stanford’s LegalBench
harvey-labs - large range of legal work tasks
prinzbench - legal research & analysis

Journals and papers

Anything by Holli Sargeant.

Ferraris, Andrea Filippo et al. “The architecture of language: Understanding the mechanics behind LLMs.” Cambridge Forum on AI: Law and Governance 1 (2025): e11
Bean, A, Payne, R, Parsons, G. et al. “Reliability of LLMs as medical assistants for the general public: a randomized preregistered study.” Nat Med 32, 609–615 (2026) – I’d think these results would apply in the legal domain
Panattoni, Beatrice. “Generative AI and criminal law.” Cambridge Forum on AI: Law and Governance 1 (2025): e9
Colombo, E. “Lifting the Veil of the Future: Arendt and Justice-Centred Artificial Intelligence.” Cambridge Yearbook of European Legal Studies (2026): 1–30
Savin, A. “When the Law Meets the Unknown: Risk, Uncertainty and the Limits of Regulation.” European Journal of Risk Regulation (2026): 1–15
Hoskins, Andrew. “AI and memory.” Memory, Mind & Media 3 (2024): e18
Kobi Hackenburg, et al. “AI systems out-persuade expert humans.” arXiv, 2026
Hao, Qianyue et al. “Artificial intelligence tools expand scientists’ impact but contract science’s focus.” Nature 649 (2026): 8099
Guien, Laura & Hoskins, Andrew. “We are an input, now. Generative AI and the death of the journalist as human memory-maker.” Memory, Mind & Media 5 (2026): e19
“Law and Empathy in the Automated State.” Cambridge University Press (2023): 173
“Supervising Automated Decisions.” Cambridge University Press (2023): 205
“Against Procedural Fetishism in the Automated State.” Cambridge University Press (2023): 221
Greene, Christopher et al. “It’s all about respect: How explanations impact trust and procedural justice in AI-assisted judicial decision-making.” Cambridge Forum on AI: Law and Governance 2 (2026): e21
Mehmood, Sultan et al. “Courts of Tomorrow: Evidence from a Nationwide Rollout of Generative AI.” CEPR Discussion Paper 21783 (2026) – randomised trial of an AI assistant across Pakistan’s trial courts
Palod, Vardhan et al. “Evaluating the False Trust Engendered by LLM Explanations.” arXiv, 2026
Hadfield, Gillian, Hendrycks, Dan & Wu, Leo. “We Need Better Infrastructure to Govern AI Agents.” AI Frontiers, 2026
Abbott, Ryan & Sarch, Alexander. “Punishing Artificial Intelligence: Legal Fiction or Science Fiction.” In Legal Aspects of Autonomous Systems, Springer (2024)

Books

These are available as Open Access.

Branting, L. Karl, and Mia Bonardi. Equalizing Justice: Harnessing AI for Litigants Without Lawyers, Cambridge University Press, 2026.
McKay, Carolyn, Digital Vulnerability in Criminal Justice: Vulnerable People and Communication Technologies, Palgrave Pivot, 2026.
Smuha, Nathalie A., editor. The Cambridge Handbook of the Law, Ethics and Policy of Artificial Intelligence, Cambridge University Press, 2025.
Gless, Sabine, and Helena Whalen-Bridge, eds. Human–Robot Interaction in Law and Its Narratives: Legal Blame, Procedure, and Criminal Law, Cambridge University Press, 2024.
Engstrom, David Freeman, editor. Legal Tech and the Future of Civil Justice. Cambridge University Press, 2023.
Voeneky, Silja, et al., editors. The Cambridge Handbook of Responsible Artificial Intelligence: Interdisciplinary Perspectives. Cambridge University Press, 2022.
Micklitz, Hans-W., et al., editors. Constitutional Challenges in the Algorithmic Society, Cambridge University Press, 2021.

People

Law and legal tech

Stanford Law School - legal informatics and AI
Justin Curl - Anthropic, future of law researcher
Damien Charlotin - creator of hallucination database
David Anthony Colarusso - legal technologist
James Grimmelmann - legal scholar: copyright, code, AI
Matthew Butterick - litigator
Niccolò Ridi - legal NLP researcher
Scott Young - legal tech commentator
Tris Sherliker - solicitor building legal tools
Umar Butler - creator of open-source legal AI benchmarks
prinz - lawyer and AI commentator, creator of prinzbench

AI governance and safety

Ajeya Cotra - future of AI and AI risk
Daniel Kokotajlo - future of AI and AI governance
Dean W. Ball - AI policy, law, and governance
Markus Anderljung - AI governance
Peter Wildeford - forecasting AI trends and AI risk

General AI commentary

Jack Clark - Anthropic cofounder
Simon Willison - software developer tracking AI capabilities and security, also see his 2026 AI summary
Zvi Mowshowitz - AI zeitgeist

Events and News

Conferences

AAAI - Association for the Advancement of AI - 16-23 February 2027 (Montreal, Canada)
AIES - AAAI/ACM Conference on AI, Ethics, & Society - 2027 dates TBA (usually October)
CodeX FutureLaw - Stanford - 2027 dates TBA (usually April, Stanford, USA)
ECAI - European Conference on AI - 2-7 October 2027 (Athens, Greece)
Geneva AI Summit 2027 - 21-22 June 2027 following Bletchley Park (2023), Seoul (2024), Paris (2025), New Delhi (2026)
IAAIL - International Association for AI & Law - ICAIL 2027 dates TBA (usually June)
IASEAI - International Association for Safe & Ethical AI - 9-12 February 2027 (Paris, France) - also see their affiliates page for a repository of supporting orgs
ICLR - International Conference on Learning Representations - 26-30 April 2027 (location TBA)
ICML - International Conference on Machine Learning - 2027 dates TBA (Rio de Janeiro, Brazil)
IJCAI - International Joint Conference on AI - 7-13 August 2027 (Kyoto, Japan), 15-17 August 2027 (Hengqin, China)
JURIX - International Conference on Legal Knowledge & Information Systems - 8-10 December 2026 (Toulouse, France)
NeurIPS - Conference on Neural Information Processing Systems - 6-12 December 2026 (Sydney, Australia)
World AI Conference - 2027 dates TBA (usually July, Shanghai, China)

News

Financial Times | Artificial intelligence
Transformer News
Lawfare
The New Yorker - Is A.I. Above the Law? (Jill Lepore)
AI Safety at the Frontier - monthly summary of AI safety papers