By Charlie Kerrigan

“They [America and China] both see the domination of AI as the key to dominating the rest of the 21st century” from The Economist, March 7^th-13^th 2026.

Since the publication of the first edition of Artificial Intelligence: Law and Regulation in early 2022 a revolution has taken place.

The elements of this revolution are each are revolutionary in themselves, and together have a compounding effect:

• The capabilities of AI systems – where do we start… Olympiad-level mathematical reasoning with full proofs; perusing a 500k word legal textbook in less than 3 seconds; “nearly half of all [Google’s] code generated by AI.” per Alphabet CFO Anat Ashkenazi in Alphabet Q3 2025 earnings call
• The accessibility of those systems – each of us can now interact with, instruct, and manage AI systems throughout our working and personal lives – the first edition of the book predated ChatGPT’s public release – you had to be a computer scientist to interact with AI – now you can talk to it and tell it what you want
• The dawning of our need to deal with the social and political implications of those first two elements of the revolution – more jobs? less jobs? who does the jobs? what are they and where?

We have vast amounts of usable, useful intelligence available to all of us and we are just starting to understand what we can do with that.

As I was completing work on the second edition, my feeds were full of people getting AI agents and teams of AI agents to build and reflexively improve what used to be entire SaaS enterprise businesses.  As Simon Taylor, the influential fintech writer, puts it in The SaaSpocalypse the week AI killed software (9 February): “This isn’t ‘AI helping a developer.’ This is a development team made of software.”

And more than that.  OpenAI’s GPT-5.3-Codex is the first model that was instrumental in creating itself.  OpenAI used Codex to debug its own training runs.

Like all revolutions, this one is not happening in the same way everywhere. Just as the French revolution felt different to a Breton fisherman than it did to a Parisian noble, the AI revolutions are happening in Palo Alto in ways they are not happening in Doncaster in South Yorkshire (where I’m from). But the difference is not just geographical. It is defined by access, attitude, and agency, as well as the ability of the users.  My 4 As of AI.

The main aim of the book is to be one of the correctives to that.  Everyone can and should engage with AI technologies.  They are “general purpose” technologies, meaning they can be used for an infinite range of tasks.  They are amazing, meaning that their capabilities are constantly surprising. They are also the most powerful tools that hackers and criminals have ever had.  We now have systems that outperform the best human hackers, that can flawlessly impersonate your family or friends, and that know everything about you. There are many aspects of AI that the world isn’t ready for.

The characteristics of AI are what makes them hard to handle, including for lawyers using and advising users.  AI is, by definition, self-learning: it improves itself autonomously.  It is, by definition, opaque: outputs are “inferred” from inputs.  The maths is not arithmetic, so outputs are not predictable.  With an algorithm, if you know all the inputs and the algorithm, you know the output. This is not how AI works.

The book is set up in a way that is intended to help people understand how AI relates to them.  It starts with explanations of AI from a few different perspectives, including technical explanations that get more challenging but don’t require deep mathematical knowledge.  It has chapters mapping AI against traditional legal topics: contract, regulation, IP, employment, corporate governance, etc.  It has chapters on AI in different industries: finance, telecoms, real estate, retail.  It has chapters on human AI: bias, security, misinformation, ethics, etc.  Finally, it has chapters on technical and consulting themes: risk management, business models, explainable AI, AI in law firms, etc.

What’s new in AI?

Generative AI is everywhere and it has moved from being a chat interface into something that does work across tools and systems.

LLMs were a new interface when ChatGPT launched to public use in November 2022.  Now AI makes the interface less important because it comprises systems that influence environments – the value is the outcome.  

Transparency has moved on. With AI agents the question isn’t just: “why this prediction?” but “why this sequence of actions, on this data, with these permissions?” Explainable AI and audit trails have turned into product requirements.

There will continue to be many new things in AI. The book aims to help define how lawyers can best engage with this evolution. Lawyers in AI aren’t adoption blockers; lawyers’ know-how in privacy, security, reliability, explainability, and governance unlock AI.

AI lawyers are the difference between demos and deployable systems.

Artificial Intelligence: Law and Regulation, Second Edition Edited by Charles Kerrigan is available to read as a Hardback and eBook. Learn more.

Free chapter available on Elgaronline.