SRA warns over AI misuse following rise in blunders

The Solicitors' Regulation Authority (SRA) warned solicitors against lax use of generative AI, citing risks of hallucinated case law and confidentiality breaches, after receiving 42 reports of potential AI misuse between July 2025 and July 2026. The SRA emphasized that solicitors remain accountable for their work regardless of AI use and that client information should only be entered into AI systems with appropriate safeguards. SRA executive director Aileen Armstrong stated that firms must have oversight and controls to ensure high-quality legal services.

Calls for appropriate oversight The Solicitors’ Regulation Authority SRA has warned solicitors against lax use of generative AI, following mounting instances of hallucinated case law and confidentiality breaches. The notice http:// comes after a string of headline-grabbing AI blunders at law firms, including Pinsent Masons https://www.legalcheek.com/firm/pinsent-masons/ ‘ citation of an insolvency rule that didn’t exist https://www.legalcheek.com/2026/05/pinsent-masons-criticised-by-judge-over-junior-lawyers-ai-blunder/ , fuelled by a junior lawyer’s AI-hallucinated research, and a similar error made by US firm Sullivan & Cromwell https://www.legalcheek.com/firm/sullivan-cromwell/ , which apologised to a New York judge after AI-hallucinated citations were found in a court filing it had submitted https://www.legalcheek.com/2026/04/sullivan-cromwell-apologises-after-ai-hallucinations-appear-in-court-document/ . While the SRA acknowledges the value AI can bring to legal services when used correctly, they stressed that “it also presents risks that must be understood and managed.” These risks centred around AI’s tendency to hallucinate as well as on potential breaches of client confidentiality when uploading sensitive materials to AI. The SRA itself received 42 reports relating to potential AI misuse between July 2025 and July 2026. On hallucinations, the notice reminds solicitors that they “remain accountable for their work and outputs, regardless of how that work has been prepared”. It also stresses that “firms must also have effective governance structures, systems and controls in place to manage any risks, including those arising from use of AI”. Turning to client confidentiality, the SRA notes that entering confidential information into open source AI tools, such as ChatGPT, would amount to a waiver of privilege, instead confirming that “client information should only be entered into AI systems where appropriate contractual, technical and organisational safeguards are in place to protect confidentiality.” The notice goes on to emphasise that it’s “incumbent on all managers and employees of SRA regulated firms to understand their obligations in this regard given the irreparable consequences of sharing clients’ confidential information and the permanent waiver of privilege that can occur as a result.” Commenting on the notice, SRA executive director strategy and policy, Aileen Armstrong said: “Firms and solicitors must have appropriate oversight and controls in place to make sure consumers are receiving high-quality legal services and that our standards are met. Individuals remain responsible for the work they produce and the advice they provide, regardless of whether AI is used, or not.” The SRA joins the Bar Standards Board BSB , which earlier this year warned barristers https://www.legalcheek.com/2026/05/hands-off-chatgpt-bsb-puts-barristers-on-notice-over-free-ai-tools/ that free AI tools would generally be unsuitable for legal work.