Law firms were early adopters of generative AI, and early casualties. The rules that apply are not new. They are the core professional duties, applied to a new tool.
United Kingdom
In R (Ayinde) v Haringey in 2025, the Divisional Court warned that lawyers who put fictitious authorities before the court face regulatory and even criminal consequences, and that responsibility cannot be passed to the tool. Further cases followed in 2026. In August 2026 the Solicitors Regulation Authority issued a warning notice to all solicitors and firms on two risks: false citations generated by AI, and breaches of client confidentiality through public AI tools. Solicitors remain personally accountable for AI-assisted work.
United States
The American Bar Association’s Formal Opinion 512 of 2024 applies the Model Rules to generative AI: competence, including understanding a tool’s limits; confidentiality, including informed client consent before inputting confidential information into some tools; supervision; candour to the court; and reasonable fees. Courts have sanctioned lawyers for fake citations since Mata v Avianca in 2023, and many judges now require disclosure of AI use in filings.
The practical risks
Confidentiality. Consumer AI tools may use inputs for training. Use business or enterprise tiers with contractual protections, and check where data goes.
Accuracy. Every citation and quotation must be checked against the source.
Billing. Charging hours for work the tool did in minutes is a fee question, not a technology one.
AI can draft. Only a lawyer can sign.
What to do
Adopt a written AI policy, approve specific tools, train everyone, and make citation checking a mandatory step before anything leaves the firm.
This article is general information on the law as at November 2026. It is not legal advice and has not been updated since that date. See the terms of use.