AI in Legal Practice

Artificial Intelligence (“AI”) has moved to the forefront of legal practice in recent times. As with any fast-developing technology, the risks and cautionary tales often attract the most attention, but they should not obscure the genuine opportunities AI presents. For lawyers, the question is no longer whether AI will have a role in legal practice, but how it can be used responsibly, effectively and consistently with professional obligations.

There is a broad consensus that AI can bring significant benefits to legal practice. Used appropriately, it can help lawyers deliver real gains for clients in quality, efficiency and cost-effectiveness. Those who strike the right balance between using AI responsibly and maintaining high legal standards will be best placed to succeed in a rapidly changing legal landscape. Equally, firms that fail to engage with AI risk being left behind.

From a litigation perspective, tasks such as document review, drafting and legal research have traditionally been time consuming. AI can now provide significant support with those tasks by assisting with the initial heavy lifting in minutes. That, in turn, gives lawyers more space to scrutinise the output, apply their professional judgement and focus on legal strategy, while helping to deliver a more cost-effective service for clients.

Although AI has the potential to reshape the legal profession, it also presents significant risks, particularly in relation to accuracy, confidentiality and professional responsibility. These are fundamental considerations for lawyers. In June 2025, the President of the King’s Bench Division ruled on two UK cases in which lawyers had submitted written arguments containing fictitious, AI-generated case citations. In one of those cases, a filing contained 45 case citations, 18 of which did not exist.

It would be too simplistic to treat these examples as a problem confined to junior lawyers or inadequate training. The risk is broader and can arise at any level of practice. Generative AI can produce responses that are fluent, confident and well-reasoned, regardless of whether they are accurate. That apparent confidence is precisely what increases the risk of hallucinated content finding its way into legal advice or court filings. It is also important to recognise that AI is not limited to lawyers. It is increasingly available to clients and unrepresented parties, who may not be subject to the same professional duties or safeguards.

The Code of Conduct of the Solicitors Regulation Authority (“SRA”) sets out the standards of professionalism that the SRA and the public expect of lawyers authorised by the SRA to provide legal services. Relevant duties include: not misleading clients, the court or others; drawing the court’s attention to relevant cases and statutory provisions of which the lawyer is aware and which are likely to have a material effect on the outcome of the case; and providing a competent service. Those duties require lawyers to retain responsibility for the work they produce, regardless of whether AI has been used to assist with it. AI-generated output should therefore be treated as a starting point rather than an answer. It must be checked, verified and assessed by the lawyer, who must then decide, using their professional judgement, whether it should be relied upon, shared with a client or put before the court.

As AI tools continue to improve and are used in increasingly varied ways, lawyers and firms must ensure that appropriate safeguards are in place. Practical steps for law firms to consider include:

• Confirming with clients at the outset whether, and in what circumstances, AI tools may be used on their individual matters and ensuring that any agreed parameters or restrictions are recorded.

• Implementing clear AI usage policies that set out which tools are approved, what they may and may not be used for, and how their use should be recorded or disclosed to clients.

• Training all lawyers, from junior to senior, on both the capabilities and the limitations of AI, including why fluent, confident output is different from accurate output, and when AI should not be used at all.

• Putting robust supervision and verification processes in place, so that every citation, fact and figure generated with the help of AI is checked against a primary source before any document is submitted to court or advice is given to a client.

• Taking confidentiality and privilege seriously by using secure, reputable platforms designed for legal use and supported by appropriate data protection safeguards, rather than inputting confidential or privileged client information into open, public AI tools.

AI now has a clear place in legal practice, and lawyers’ reliance on it is only likely to grow. But caution remains essential. AI is a tool, not a substitute for proper research, review, supervision or professional judgement. The firms that benefit most from AI will be those that treat it not as a shortcut, but as a powerful tool to be used carefully, transparently and under proper professional oversight.