AI In Legal Practice: Innovation Must Not Come At The Expense Of Professional Judgement

How solicitors can use AI responsibly while complying with SRA requirements on accuracy, confidentiality, supervision and professional judgement.

Artificial intelligence is rapidly reshaping the legal sector. Tasks including legal research, document review, drafting and case analysis can now be undertaken more quickly with the support of AI-powered tools. User appropriately, AI offers firms an opportunity to improve efficiency, reduce administrative burdens and enhance the service delivered to clients.

However, as adoption increases, so do the professional and regulatory risks.

On 17 August 2026, the Solicitors Regulation Authority published a warning notice entitled “Misuse of AI”. While recognising the benefits that AI can bring to legal practice, the SRA’s central message is clear: using AI does not reduce, remove or transfer the professional responsibilities of solicitors and law firms. Regulated individuals remain accountable for the work and advice they provide, while firms must maintain effective governance, supervision, systems and controls.

A Growing Concern for the Profession

The SRA’s intervention follows a series of incidents involving the misuse of AI in legal services. According to the SRA, it received 42 reports related to the potential misuse of AI between July 2025 and July 2026 and has ongoing investigations relating to inaccurate legal citations, supervision and confidentiality.

Two principle concerns emerge from the warning notice:

The first is inaccurate or fabricated content generated by AI systems. Large language models can produce responses that appear authoritative and convincing even when they have no proper factual or legal basis. These so-called “hallucinations” may include invented cases, false citations, inaccurate quotations and incorrect legal propositions.

The second concern is confidentiality. Client documents, personal data or legally privileged material entered into an AI system without appropriate safeguards may be stored, retained, accessed or used in ways that compromise professional obligations. Importantly, these risks are not confined to free or publicly accessible platforms. Paid-for tools may also lack the contractual or technical safeguards required for sensitive client information.

When AI Gets the Law Wrong

The warning notice places particular emphasis on the danger of relying on AI-generated legal authorities without proper verification.

The legal profession depends on accuracy, credibility and trust. Solicitors must ensure that submissions are properly arguable and that any cited authority is genuine, relevant and supported by a verifiable citation.

In R (on the application of Ayinde) v Haringey London Borough Council [2025] EWHC 1383 (Admin), false case citations were placed before the court. The solicitor and barrister faced a wasted costs application and were referred to their respective professional regulators. The court made clear that reliance on AI would not provide a suitable defence and that lawyers remain responsible for verifying the content of their submissions.

Other decisions have reinforced that message. In BCP v A Mother [2026] EWFC 71 (B), the court treated AI-generated inaccuracies seriously even though it did not find a deliberate intention to mislead. In Cork and another v Smith [2026] EWHC 1199 (Ch), the court again emphasised the responsibilities of authorised legal professionals where errors were attributable to unchecked AI output.

The consequences may extend beyond professional embarrassment. Depending on the circumstances, false material placed before a court may result in wasted costs, regulatory referral, disciplinary action or possible contempt proceedings.

Confidentiality in the Age of AI

If inaccurate legal research represents one side of the AI risk equation, confidentiality represents the other.

Client confidentiality lies at the heart of a solicitor-client relationship. However, the way in which AI platforms process information may not always be apparent to users. Depending on the provider’s terms, settings and technical architecture, information entered into an AI system may be stored, retained, accessed by third parties or used to improve the service.

In UK v Secretary of State for the Home Department [2026] UKUT 81 (IAC), the Upper Tribunal raised concerns about placing client correspondence into an open-source AI tool. The SRA relies on that decision when warning that such conduct may breach confidentiality and may compromise or waive legal professional privilege.

Client information should therefore be entered into an AI system only where the firm is satisfied that appropriate contractual, technical and organisational safeguards are in place. This includes understanding where data is processed, who may access it, how long it is retained and whether it may be used to train or improve an AI model.

Where possible, users should also minimise the information provided and consider whether confidential or identifying details can be removed. Most importantly, they should use only tools that have been approved for the proposed purpose.

Responsibility Starts at Firm Level

Compliance is not solely a matter for individual fee earners.

Firms must have effective governance structures, systems and controls to manage the risks arising from AI use. Managers, partners, supervisors and compliance officers should understand how AI is being used within their organisations and ensure that those using it receive appropriate training and oversight.

This is particularly important because AI tools are widely and easily accessible. Employees may experiment with unapproved platforms without fully appreciating the professional risks. A clear AI policy should therefore identify which tools may be used, the purposes for which they are approved, the information that must not be entered and the level of human checking and supervision required.

Supervisors also remain accountable for work carried out through those they manage. Review processes must therefore be capable of identifying inaccurate or unsupported AI-generated material before it reaches a client, court or third party.

Balancing Innovation and Professional Responsibility

The SRA’s message is not that AI should be avoided. Used within an appropriate framework, it can support research, drafting, document review and the efficient delivery of legal services.

However, solicitors must continue to exercise independent judgement. Legal research must be verified. Authorities must be checked. Client confidentiality must be protected. Supervision must remain effective. Decisions must be capable of justification. Accountability remains with the individual and the firm, not the technology.

The firms best placed to realise AI’s benefits are likely to be those that treat it as a tool supporting, rather than replacing, professional judgement. The SRA’s warning notice is therefore not a barrier to innovation. It is a reminder that innovation in legal services must be accompanied by effective safeguards, informed oversight and continuing professional responsibility.