On 17 August 2026, the Solicitors Regulation Authority (SRA) published a warning notice (the Warning Notice), addressed to all regulated firms and individuals, regarding the misuse of AI. The Warning Notice highlights particular concerns relating to the risk of inaccurate information being included in court and other legal documents, and risks to client confidentiality.

In the Warning Notice, the SRA emphasises that law firms and individual solicitors in both private practice and in-house teams must ensure that they use AI in ways that are compliant with the SRA Codes of Conduct and their duties owed to the court. The Warning Notice states that in-house counsel should be particularly mindful about their organisations’ use of AI and the types of tools that may be available to them which may not be specifically designed for legal work.

The Warning Notice has been published in a broader context of regulators and professional organisations issuing guidance to provide clarification on how their rules apply to the use of AI. In the UK, this includes the Bar Council, the Financial Conduct Authority and the Information Commissioner’s Office.

In this article we summarise the key messages in the Warning Notice.

The SRA’s concerns

The SRA takes an outcome-based approach to regulation, which means that regulated persons can decide how to use AI, provided they comply with the SRA’s standards. The Warning Notice recognises that many solicitors and law firms use AI safely and responsibly. However, the SRA states that it has identified a number of cases involving the misuse of AI, and highlights two key areas of concern – (1) inaccurate information in court and other documents as a result of AI tools producing ‘hallucinations’; and (2) failures to consider and appropriately mitigate risks to client confidentiality when using AI tools, for example, cases where confidential client information has been entered into public AI tools. In contravention of the SRA’s standards, such misuse of AI can cause harm to clients, have a negative impact in the courts and harm the standing of the legal profession by damaging public trust and confidence.

1. AI hallucinations

AI tools can generate outputs which seem plausible but are wrong or fabricated. There have been several recent cases where solicitors (and barristers) have submitted documents to court which contain fictitious case references generated by AI which have not been verified prior to submission.

The Warning Notice notes that AI does not have legal personality and emphasises that solicitors and other regulated individuals who use AI tools in the course of their work remain accountable under the Code of Conduct for their output, regardless of how they prepare the work. Further, as officers of the court, solicitors who mislead the court risk serious consequences. The Warning Notice explains that: “To uphold your duty to the court, you should be assured that all submissions of named case law authorities are genuine, relevant, have a verifiable citation, and advance the arguments that are being put forward in your documents.” Putting false material before a court can also be considered contempt of court.

The Warning Notice also emphasises the importance of adequate and effective supervision. Those who supervise junior or non-authorised colleagues may also breach regulatory requirements and their professional duties if false citations are put before the court without adequate review and supervision. Similarly, firms are required to maintain effective systems for supervising client matters.

2. Client confidentiality

In the Warning Notice, the SRA reiterates the earlier viewof the Upper Tribunal that putting client documents into AI tools such as ChatGPT will likely breach client confidentiality and waive legal privilege. This decision of the Upper Tribunal has generated significant commentary in the legal community. As noted in a speech by the Master of the Rolls on AI and legal professional privilege, the Upper Tribunal did not examine precisely how information entered into systems such as ChatGPT are treated. Further, the decision described ChatGPT as “open source”. While it appears the Upper Tribunal used “open source” to mean public AI tools, in software terminology, “open source” means making the source code of software publicly available, which does not apply to ChatGPT: it is a proprietary AI model. Notwithstanding these points, the SRA has endorsed the principle in the decision and given a clear warning regarding the risks to client confidentiality of putting client information and documents into AI tools.

The Warning Notice explains that both free to use and paid for AI systems may pose risks to client confidentiality – depending on the provider’s terms and the AI system’s settings and technical architecture, information entered into the system may be stored, retained or used to improve the tool. Similarly, as more law firms explore the use of AI systems and tools, it is important to ensure that client confidentiality is maintained.

Many enterprise AI tools can be used consistently with client confidentiality obligations provided the appropriate safeguards are in place. The Warning Notice recognises this and highlights the need for appropriate contractual, technical and organisational safeguards. Key considerations identified by the SRA include maintaining full control over how information is processed or used, whether data remains in a secure environment, whether data can be accessed by unauthorised third parties, and whether data is used to train AI models (except if explicitly authorised and appropriate to do so).

The Warning Notice also provides further guidance for in-house lawyers, noting that AI tools developed for use by the business may not be specifically designed for legal work and therefore care must be exercised. It further states that in-house lawyers should “consider whether the organisation's interests in furthering its use of AI might conflict with any of their own duties and professional responsibilities under the Principles and Code of Conduct, dependent upon their role and use of those tools”.

Key takeaways

As the use of AI tools by law firms and in-house teams continues to grow, there will be increased focus from the SRA on ensuring such tools are used in compliance with professional obligations and duties owed to the courts.

The Warning Notice makes clear that AI tools can be valuable when used correctly, but there are significant risks which must be understood and managed. Regulated persons remain accountable for the work they produce, for maintaining confidentiality and complying with their duties to the court. AI tools cannot diminish or transfer professional responsibilities for the quality and accuracy of services.

Given the speed at which AI is developing and being adopted, there are significant challenges for the legal sector and in-house teams to develop mature and effective governance arrangements. An approach toward AI tools which is taken in haste, without due care, or on a one-size-fits-all basis, will pose significant risks to compliance with professional obligations. Firms and in-house teams must ensure there is robust legal and technical oversight over AI systems and have appropriate governance arrangements and training in place in order to manage risks of the type identified by the SRA in the Warning Notice.

With thanks to Emma Torok for her assistance in preparing this post.

Subscribe to our Inside Disputes blog to receive regular expert insights into banking and commercial litigation in the English courts and issues relevant to investigations.

Recent articles

  • content-type:insights/blog
  • content-type:generic
  • nrf-migrated:article-categories/inside-disputes/type/blog-post