Cork & Anor v Smith

Answer

AI is already changing the way legal services are delivered. Across the market, firms are investing in tools that can speed up research, produce first drafts, review documents more quickly and reduce the time spent on routine tasks. From a commercial perspective, that’s easy to understand. Clients want work to be delivered efficiently and cost-effectively, and firms are under pressure to show they can embrace new technology without compromising quality. However, the recent decision in Cork & Anor v Smith is a reminder that the legal profession can’t look at AI purely through the lens of efficiency. The case shows that while AI may offer real opportunities for firms, it also creates clear legal, regulatory and reputational risks if used carelessly. It therefore provides a useful illustration of the way in which innovation and professional responsibility must now be considered together.

Cork & Anor v Smith: an AI hallucination

Cork & Anor v Smith arose from what should’ve been a relatively routine insolvency application. The applicants’ former solicitors wrote to the court asserting that it had an express power under a particular insolvency rule, and they set out wording said to support that proposition. The problem was that the wording relied upon didn’t exist. When the court checked the cited provision, it found that the quoted text appeared nowhere in the legislation. The judge concluded that the supposed rule was an AI hallucination: false material generated by AI and presented as though it were authoritative legal text.

The seriousness of the case lay not only in the original error, but also in the response that followed. When the court queried the point, a further response failed to provide a credible explanation. The judge criticised the approach taken to the accuracy of the material put before the court and described himself as astonished by the follow-up. The matter was self-reported to the Solicitors Regulation Authority, underlining the wider professional consequences. The court’s message was straightforward: AI may be a useful starting point, but it can’t replace legal analysis or the need to verify material before it’s deployed in litigation.

Why the case matters

Cork & Anor v Smith matters because it captures a wider tension in the legal market. AI is attractive to firms because it promises faster turnaround times, lower costs and greater efficiency in routine work. Those pressures are real in a market shaped by client demand, intense competition and growing expectations around the use of legal technology. At the same time, the case shows why speed can’t be prioritised over reliability. A fabricated authority or misleading legal proposition can affect a client’s position, waste court time and damage a firm’s credibility. In commercial terms, the risks are therefore not confined to legal error; they extend to reputation, client confidence and regulatory exposure.

The case is particularly important in a legal context because legal work depends on authority, accuracy and professional judgment in a way that many other sectors don’t. A misleading sentence in an internal business document may be inconvenient; a false authority in litigation can misdirect the court, undermine a client’s case and create costly consequences. Legal analysis is also highly jurisdiction-specific, which makes AI-generated generalisations especially risky. A response that appears plausible on its face may still be wrong in substance, outdated in law or unsupported by any real source. That makes legal work especially vulnerable to the confident tone of generative AI. In practice, the danger is not only that AI may invent material, but also that it may present that material in a way that looks polished enough to escape immediate challenge.

The courts’ message on AI and professional responsibility

Cork & Anor v Smith isn’t an isolated example. In R (Ayinde) v London Borough of Haringey, the court stressed that lawyers who use AI for legal research must still check the accuracy of that research against authoritative sources before relying on it. In A, B, C, D, Re (Extension of assessment; Use of AI: hallucinations), the court again had to deal with submissions containing authorities or propositions that weren’t real. Taken together, these cases show an emerging judicial approach: AI has a place in legal practice, but its use doesn’t lower the standard expected of lawyers. If anything, it increases the need for supervision, sound judgment and attention to detail.

What this means for firms

The judicial approach is reinforced by the wider guidance issued by the Courts and Tribunals Judiciary, which recognises that AI is likely to play an increasing role in legal and judicial work but emphasises the associated risks, including hallucinations, inaccurate information, bias and confidentiality concerns. Crucially, the guidance makes clear that responsibility remains with the person using the technology. AI may improve processes, but it doesn’t transfer accountability.

For firms, the key question is therefore not whether AI should be used at all, but how it can be used safely and responsibly. The potential benefits are too significant to ignore, particularly in areas such as productivity, turnaround times and cost management. However, the firms most likely to benefit will be those with clear governance around its use: approved tools, internal policies, proper supervision and a culture in which AI output is treated as a starting point rather than a final answer. In practice, AI isn’t just a technology issue. It’s also a risk, training and client service issue.

The response to AI risk is likely to become more formalised. The Civil Justice Council’s consultation paper on the use of AI for preparing court documents reflects a broader recognition that existing professional duties may need to be supported by clearer procedural expectations in practice. That may include stronger requirements around disclosure of AI use in certain contexts, more structured training for legal representatives and clearer firm-wide policies on when public or private tools may be used. Confidentiality will remain a central issue. Even where the output is accurate, entering non-public client or case information into an unapproved system may create a separate risk entirely. Therefore, for firms responsible AI adoption isn’t just about checking authorities after the event. It also requires decisions about procurement, supervision, data security and internal governance from the outset.

Ultimately, the case highlights a simple but important point: persuasive wording isn’t the same as accurate legal analysis. One of the main risks of generative AI is that it can produce fluent and confident text that appears reliable even when it’s wrong. That makes verification essential. Clients may value efficiency, but they also expect judgment, reliability and accountability. Cork & Anor v Smith suggests that firms that gain the greatest advantage from AI will not be those that use it most aggressively, but those that use it with the strongest controls.

The direction of travel is clearly towards greater use of AI in legal practice. But Cork & Anor v Smith shows that any commercial advantage will depend on how responsibly that technology is used. The firms best placed to adapt will be those that combine innovation with effective oversight, recognising that while AI may assist with speed and efficiency, responsibility for the final product still rests with the lawyers.

Olivia Gardiner is a trainee solicitor at DWF Group Limited.