Artificial intelligence did not enter the legal profession with a policy decision or regulatory approval. It arrived quietly through research tools, drafting assistants, document review platforms and, increasingly, through lawyers’ own personal accounts used under the deadline pressure of their transactional work.
That is why recent judicial warnings about fabricated citations feel both shocking and familiar. They are not evidence of a future problem. Rather, they are evidence that AI is already embedded in legal work, but unevenly governed and imperfectly understood.
This growing disconnect between how AI is actually being used and how our professional frameworks assume it is being used represents two sides of the coin: the profession’s greatest risk and greatest opportunity.
From shadow use to structured capability
Much commentary still frames AI in law as experimental. In reality, many firms have moved well beyond experimentation. Specialist legal platforms, from Harvey and Luminance to Legora and AI-enabled tools from LexisNexis, are now embedded in research, drafting, due diligence and knowledge workflows across parts of the market.
Less visibly, AI has become ambient. Familiar tools such as Microsoft Teams and Zoom now offer transcription and summarisation as standard, while Microsoft Copilot is being woven into Microsoft Word, the profession’s perennial tool of choice. In many firms, AI already operates quietly in the background of routine legal work, whether it is recognised as such or not.
What is striking is not simply the speed of adoption, but the pattern. Where firms invest in training, supervision and workflow integration, AI use becomes normalised, auditable and measurably beneficial. Where they do not, usage still happens but informally, inconsistently and outside firm controls.
This is the uncomfortable truth behind so-called “shadow AI”. It is not driven by recklessness, but by capability gaps. Lawyers are solving immediate problems with whatever tools are available. Blanket bans do not stop this behaviour; they merely push it out of sight.
Quality, not shortcuts
There is a persistent fear that AI lowers standards. Used badly, it can. Used well, the evidence increasingly points the other way.
In firms that have moved beyond pilots, AI is not replacing legal judgment; it is supporting it. Lawyers use it to surface missing issues, stress-test reasoning, compare approaches across jurisdictions and accelerate drafts that are then properly reviewed. Junior lawyers, in particular, benefit when AI is used as a structured assistant rather than an unsupervised shortcut.
The real differentiator is not the tool, but how it is used. Firms that treat AI as a skill (i.e. something to be learned, supervised and improved) see compounding benefits and value over time. Those that treat it as a gadget do not.
This matters because the future of the profession is not simply about doing the same work faster. Increasingly, firms are using AI to rethink how legal services are delivered altogether – from new pricing models to more integrated, problem-solving approaches that recognise clients are rarely bringing “legal problems” in isolation, but rather business challenges in which legal issues are only one component.
The risk conversation we are not having
Opportunity, however, does not cancel risk. It sharpens it.
From the perspective of professional indemnity insurers, AI is not a novelty. It is an amplifier. Errors can scale more quickly; provenance becomes harder to evidence; supervision models designed for traditional workflows come under strain. Cyber insurers, meanwhile, are increasingly focused on how AI tools interact with data security, access controls and confidentiality.
Yet many firms still discuss AI primarily as a productivity initiative, rather than as a cross-cutting professional risk issue that spans supervision duties, insurance disclosures, cyber resilience and client trust.
This is where regulatory posture matters. The Law Society of England and Wales has rightly emphasised the principles of competence, confidentiality and professional judgment. But principles alone do not answer the operational questions firms are now grappling with: what training is sufficient, what supervision looks like in AI-assisted work, and how firms demonstrate control rather than aspiration.
On those questions, the regulatory response, particularly from the Solicitors Regulation Authority, is yet to catch up with practice. High-profile announcements of AI-enabled services, or the creation of regulatory sandboxes, can signal intent, but they do not by themselves establish consistent expectations or practical market standards. Without clearer engagement on how AI should be governed in day-to-day legal work, firms are left to interpret the rules for themselves, with uneven results.
Realism beats restraint
There is a lingering belief that caution equals safety. In the AI context, that is increasingly misplaced.
“Wait and see” has not reduced AI use in law firms. It has simply delayed the investment in governance, training and data foundations that make AI use safer and more valuable. The law firms pulling ahead are not the most reckless adopters; they are the most deliberate ones.
They are clear about what AI can and cannot be used for. They train their people, they embed tools into approved workflows, they keep humans firmly in the loop and crucially, they can explain to clients, insurers and regulators how AI fits within their professional obligations. And that is what maturity looks like.
Ultimately, AI will not replace lawyers. But it is already reshaping how legal work is produced, reviewed and delivered. The question is whether the profession chooses to treat this as an external threat to be constrained, or as an internal capability to be governed and improved.
Handled properly, AI offers the chance to reduce drudgery, improve consistency, support junior development and strengthen client outcomes, all whilst preserving the core values that define legal professionalism.
A profession at a crossroads
The choice is not whether to engage with AI. That choice has already been made quietly, unevenly and most importantly, irreversibly. The real choice now is whether we confront that reality honestly, and build the governance, skills and supervision models that allow the profession to emerge stronger rather than exposed.
For a UK legal services sector that is a critical contributor to national growth and global competitiveness, getting this right is not optional. It is an opportunity and a responsibility to lead.
