AI and the Role of the Modern Law Library
It is difficult to think of another technology that has moved so quickly from novelty to necessity as generative AI. When ChatGPT 3.5 appeared in November 2022, it transformed public access to large language models by placing sophisticated AI behind a free and easy to use conversational interface. In the years since, generative AI has evolved from a curiosity to the commonplace, embedding itself into the workflows of many different professions, law being no exception.
For law libraries, the significance of AI extends beyond the introduction of just another research tool. AI is transforming the way legal information is created, discovered and consumed, while simultaneously making questions of trust and verification more important than ever. As information becomes easier to generate but harder to verify, the modern law library has an increasingly important role in helping practitioners distinguish authoritative legal information from convincing but unreliable content.
The growing importance of trust and verification is inseparable from the speed with which AI has become part of everyday legal practice. A LexisNexis survey of UK lawyers found that the proportion using generative AI rose from 46% to 61% in just eight months during 2025, having stood at only 11% two years earlier. More recently, Clio’s 2026 Legal Trends Report found that 89% of legal professionals now use AI in some capacity.
Across the legal profession, AI is becoming an increasingly routine part of practice. Barristers and other legal professionals are using it to summarise lengthy documents, organise and index bundles, build chronologies, assist with legal research and support drafting, while Microsoft Copilot is being utilised across the judiciary, and organisations throughout the justice sector continue to explore a growing range of use cases. Used appropriately, these tools can deliver significant efficiency gains, reducing the time spent on routine administrative tasks, allowing practitioners to focus on legal analysis and advocacy.
At the same time, AI is raising difficult legal and ethical questions at every stage of the justice system. Police use of live facial recognition technology has prompted concerns about privacy, bias and civil liberties, while the recently announced piloting of AI legal assistants in the Crown Court has reignited debate over whether such tools might undermine the fairness of proceedings if not deployed with appropriate safeguards. Alongside these wider questions, AI-generated fabricated authorities have begun appearing in court submissions from both unrepresented litigants and legal professionals, demonstrating that rapid innovation is being accompanied by equally significant risks for the administration of justice.
Adoption, although increasing, remains uneven. Specialist legal AI platforms can be prohibitively expensive, while some practitioners consciously choose not to use generative AI at all, citing concerns ranging from reliability and professional ethics to its massive environmental impact. Yet AI is increasingly becoming part of the everyday architecture of legal work. As AI capabilities are embedded within office software, search engines and document management systems, choosing not to use generative AI no longer necessarily means avoiding it altogether.
The more routinely these tools become part of everyday research and drafting, the easier it becomes for their mistakes to pass unnoticed. Recent high-profile cases involving fabricated authorities illustrate this point.
The Divisional Court’s decision in R (Ayinde) v Haringey London Borough Council [2025] EWHC 1383 (Admin) brought the issue sharply into focus when submissions relied upon authorities that simply did not exist. Yet fabricated citations are only one manifestation of a broader problem. AI systems can misstate legal principles, omit important context or present uncertain conclusions with complete confidence. Because the output is often fluent and persuasive, users may mistake plausibility for accuracy.
That lesson was reinforced in Tobosaru v Court of Law Craiova, Romania [2026] EWHC 1720 (Admin). Non-existent authorities found their way into submissions made on behalf of the respondent and were identified only by the opposing party. The CPS accepted that while generative AI was the immediate source of the error, the underlying cause was human: the authorities had not been verified prior to them being put before the court. Framed in this way, the profession has less of a hallucination problem than a verification problem.
Adoption, although increasing, remains uneven. Specialist legal AI platforms can be prohibitively expensive, while some practitioners consciously choose not to use generative AI at all, citing concerns ranging from reliability and professional ethics to its massive environmental impact. Yet AI is increasingly becoming part of the everyday architecture of legal work.
Why is verification so hard? Automation blindness is the well-established tendency for people to place increasing trust in automated systems while checking their work less rigorously. As AI becomes more capable, users naturally become more inclined to rely upon it. Alongside this sits the twin risks of a gradual de-skilling of experienced practitioners and the risk of ‘never-skilling’ for those newly entering practice where AI shortcuts risk bypassing the development of robust legal research skills.
The case of Cork v Smith [2026] EWHC 1199 (Ch) concerns a solicitor who quoted an AI system’s output, despite its explicit warning that it could not verify the legislation it was citing. The warning was ignored. The problem was not that the technology failed to signal uncertainty, but that the human user failed to act upon it.
These cases illustrate that the principal risk is rarely solely technological. Rather, it lies in the interaction between human judgment and increasingly persuasive automated systems. Law libraries therefore have an important educational role: not discouraging the use of AI, which I would argue would be unrealistic and retrograde, but ensuring that practitioners retain the habits of verification and critical evaluation on which legal research has always depended.
If the profession’s challenge is no longer access to legal information but the ability to evaluate increasingly persuasive automated output, then the role of the modern law library must necessarily evolve alongside it.
Increasingly, we are asked not simply which databases to use, but which AI tools can be trusted, when they are appropriate, how their outputs should be verified and by what means. These are questions relating to authority, provenance, evidence and professional judgment – the very foundations of legal information work.
In many respects, this is familiar territory. Law librarians have guided practitioners through successive waves of technological change, each transforming how legal information is accessed without diminishing the need to evaluate its authority. Generative AI represents the latest stage in that evolution.
Our role is therefore expanding from supporting legal research to supporting AI-enabled legal research. That means helping users understand both the capabilities and limitations of these systems, encouraging responsible experimentation and reinforcing the habits of verification that remain essential regardless of how information is generated.
Education has become one of the defining responsibilities of the modern law library. Experienced practitioners need support in recognising the limitations of AI-assisted research, while those entering the profession must continue to develop the underlying legal research skills that AI cannot completely replace. Constructing effective searches, tracing legislative amendments, checking authorities and evaluating competing sources remain core professional competencies. AI should augment those skills, not substitute for them.
Professional bodies and regulators have reached much the same conclusion. Although their guidance differs in emphasis, the underlying message is consistent: AI can be a valuable professional tool, but it does not alter the duties of competence, diligence or independent judgment that already govern legal practice.
AI will undoubtedly continue to evolve, becoming more capable, more accessible and more deeply integrated into legal practice. The question is no longer whether AI has a place in the profession; it already does. The more important question is whether legal professionals continue to develop the judgment needed to use it effectively and responsibly.
That challenge places the modern law library at the centre rather than the margins of legal practice. In an age when information can be generated instantly but not necessarily trusted, our role is not simply to provide access to legal information. It is to help practitioners distinguish authoritative information from persuasive imitation, fostering the habits of critical evaluation on which the administration of justice depends.
For members of The Inner Temple, that means an open invitation to experiment with these tools, explore their potential, and bring us your questions, concerns and the occasional AI-generated puzzle to untangle. The conversation about AI in legal practice is only just beginning, and the modern law library has an essential role in shaping how that conversation develops.
Sally Mclaren
Reader Services Librarian