About
Safe AI in Law helps lawyers build capability in advanced AI safety.
What we do
Safe AI in Law helps legal professionals recognise and assess advanced AI risk.
We translate technical knowledge into resources, regular updates and continuing professional development (CPD). Distribution through legal associations and professional programmes makes the material accessible and signals its practical relevance. An initial learning pathway builds practical understanding of AI models, systems, training, evaluations, agents, misuse and loss-of-control risks.
This also prepares lawyers to handle other pressures, such as a court facing a substantial increase in AI-generated filings. When institutions act on informed advice, they can improve safeguards, oversight and response capacity.
Incorporating such education into professional education is the ambitious goal. Regulated professions like law require practitioners to do a certain amount of continuing education each year. Having regulators adopt AI safety as part of the mandated curriculum would ensure all lawyers in a given jurisdiction do AI safety training each year.
We also send a free monthly briefing on developments, research and regulation. You can subscribe here.
Why lawyers
Lawyers already help shape the rules, incentives, evidence, and decisions that determine how the risk is handled. Yet AI risk remains absent from most legal training and continuing professional development.
Lawyers advise, draft, litigate and work in the institutions that shape how AI is developed and used. Lack of accessible and practically relevant AI safety education limits their ability to recognise and respond to serious risks. Equipping practitioners in their existing roles offers a route to improve decisions, and mitigate AI related risk, across an array of domains, for example:
- Transactional & corporate lawyers control compute purchase and model/system deployment agreements, indemnity clauses, and supply-chain liability.
- In-house governance & compliance lawyers sit between engineering teams and enterprise boards.
- Litigators & trial advocates help shape precedent.
- Regulators and policy drafters help formulate statutory definitions and oversight.
Who runs it
Martin Radzaj is a barrister with over a decade’s experience across law and government policy. He has supervised lawyers, sat on the board of several non-profit organisations, spoken at parliamentary inquiries, and taught law to students, lawyers and other barristers. Alongside his practice, Martin is pursuing technical AI safety research bridging mechanistic interpretability and law. Martin believes the legal community has a vital role to play in ensuring safe and fair AI.
Shahil Goodka is a technology & regulatory lawyer (English-law qualified solicitor) from London, most recently an Associate in RPC LLP’s leading commercial & technology practice. He was previously a Secondee to Meta’s legal team, advising on global cybersecurity incidents and data governance. Shahil now provides legal analysis for AI policy research as a SPAR Research Fellow and advises AI safety non-profits.
Get in touch
Email [email protected].