The Regulatory Context: 42 Reports and Active Investigations
On 17 August 2026, the Solicitors Regulation Authority issued a formal Warning Notice titled "Misuse of AI". Unlike informal compliance tips or research outlooks, a warning notice is a formal regulatory instrument that the SRA explicitly takes into account when exercising disciplinary functions.
The regulatory backdrop is immediate. In the accompanying announcement, the SRA disclosed that it had received 42 reports related to the potential misuse of AI between July 2025 and July 2026. More critically, the SRA confirmed that it has open, ongoing investigations relating to:
- Inaccurate legal citations and hallucinated authorities presented to courts and transaction counterparties.
- Failures of active supervision where partners rubber-stamped unverified AI drafts.
- Breaches of client confidentiality resulting from unapproved third-party tool usage.
Crucially, reports have originated not only from opposing counsel or self-reporting solicitors, but directly from senior members of the judiciary.
The Core Provisions Cited by the SRA
The Warning Notice restates existing Code obligations with sharp enforcement emphasis. For conveyancing practices, the relevant provisions form the operative audit test:
- Code of Conduct for Solicitors, Paragraph 3.5: You must effectively supervise work being done for clients. Delegating a draft to an AI system does not dilute this duty.
- Code of Conduct for Solicitors, Paragraph 7.2: You must be able to justify your decisions and actions in order to demonstrate compliance with your regulatory obligations. If asked how an AI output was verified, an unrecorded check fails this test.
- Code of Conduct for Firms, Paragraph 2.1: Firms must have effective governance structures, systems, and controls in place to manage any risks, including those arising from the use of AI.
- Code of Conduct for Firms, Paragraph 9.1: A firm's Compliance Officer for Legal Practice (COLP) must take all reasonable steps to ensure compliance with relevant regulatory arrangements, including those relating to supervision.
Language Discipline: "Appropriate" vs "Meaningful" Oversight
Precision matters when framing legal ops. The SRA Warning Notice requires "appropriate human oversight, informed professional judgement and a proportionate, risk-based approach."
Firms frequently borrow the phrase "meaningful human oversight" from Article 14 of the European Union AI Act or general UK GDPR data protection doctrine. In UK legal regulation, the SRA's governing term is appropriate. Appropriate oversight is proportionate to transaction risk: a routine client welcome email carries a different supervisory requirement than a complex leasehold covenant analysis or a Report on Title.
The Three Layers of Evidenced Oversight
UtterConnection translates the SRA's expectations into three operational layers, modelled on the standards adopted by large national practices but sized specifically for independent conveyancing firms:
1. Human On the Loop (Governance)
Which AI tools are authorised for use, on what specific matter types, with what data boundaries, and signed off by whom. Held in a live register owned by the COLP. Zero unvetted consumer tools.
2. Human In the Loop (Task-Level Supervision)
Operationalised via the Marked Draft Standard. Every AI-assisted draft reaches the reviewer explicitly labelled as AI-assisted, with the prompting objective declared and at least one open query identified. This eliminates nominal "click-through" review.
3. Human Over the Loop (Ultimate Accountability)
In accordance with the SRA's Effective Supervision guidance (12 June 2026), an authorised individual must retain ultimate responsibility for all legal services delivered. A named practitioner signs and is recorded in the case management audit log.
Moving from Symbolic to Operational Maturity
In Practical AI Governance (Kogan Page, 2026), Rosenberg and Polonetsky distinguish between Symbolic governance (policies that exist on paper but have untested controls) and Operational governance (contemporaneous records that survive forensic scrutiny). Most conveyancing practices remain Symbolic: they have a written policy in a drawer, yet their conveyancers routinely use shadow AI or unlogged prompts.
When an SRA investigator or a Professional Indemnity insurer examines a conveyancing firm, they will not ask whether the firm has an AI policy. They will ask to see the supervisory log for matter #40291. Operational governance provides that proof automatically.
Audit Your Firm's SRA Supervision Readiness
Discover whether your current fee earner workflows generate auditable supervisory records or expose the COLP to personal liability.
Take the Free AI Reality Check