← All guidelines·Guideline 3 of 8
Counsel who use AI must be reasonably competent in its use: they know what the tool can and cannot do, what happens to what is put into it, and how it fails, and they guard against those risks.
Red · Must Proposed addition. The Code does not state this duty separately; it follows from the duty of competence.
The rules
“use their best efforts to carry out work in a competent and timely manner and not take on work which they do not reasonably believe they will be able to carry out in that manner;”
LPC Code of Conduct, para 3.11
“remain reasonably abreast of legal developments, applicable laws and regulations, legal theory and the common law, and legal practice in the fields in which they practise;”
LPC Code of Conduct, para 3.13
For many advocates the use of AI is now part of legal practice. Competence in the work includes competence in the tools chosen to do it.
What counsel needs to know about each tool
- How it gets things wrong. It can produce content that is inaccurate, incomplete or invented, and present it with confidence.
- What happens to what is put in. Where the material is stored, for how long, who can see it, and whether it is used to train the model.
- What its settings do. In particular, the settings that affect confidentiality.
- What it can and cannot read. Scanned documents, handwriting and watermarks are common points of failure.
This applies to legal-specific tools and to AI functions built into research databases, not only to general chatbots. In Northbound, the incorrect citations came from a subscription tool that claimed to be trained exclusively on South African judgments and legislation (at [89]).
What is reasonable depends on the use and on the risk that use creates for the client and the court. A tool used to reformat a list calls for less than a tool used to research the law for heads of argument.
Green · Method The checks
- Read the terms. For the plan and account actually in use, not the provider’s marketing.
- Check the settings. Training, retention, sharing and connected applications.
- Test the tool on known work. Ask it about a judgment counsel knows well and note what it gets wrong.
- Keep up. Tools change often. Re-check when the tool or the plan changes.
- Learn from near misses. When the tool gets something wrong, record how, and adjust the checking routine.
Guide 1 sets out an hour of tests that show what a tool gets wrong before it is trusted, and Guide 5 describes what to read before a plugin is allowed near a matter.
Other jurisdictions
In England and Wales, barristers should maintain a sufficient level of competence in technology and AI to understand how they may affect their practice, whether or not they adopt the technology themselves (Bar Standards Board, Guidance on the use of AI and other technologies, valid from 18 May 2026, Core Duty 7). In Ontario, licensees are told to take the time to understand how generative AI works, including its capabilities, limitations and terms of use (Law Society of Ontario, Generative AI: your professional obligations, practice tip 2). In New Zealand, improper, negligent or incompetent use of generative AI could lead to a serious breach of the conduct rules (New Zealand Law Society, Lawyers and generative AI, March 2024). The international comparison has the detail.
Sources on file: LPC Code of Conduct (GG 42337, 29 March 2019), paras 3.11 and 3.13. Northbound Processing (Pty) Ltd v South African Diamond and Precious Metals Regulator [2025] ZAGPJHC 661, para [89]. Bar Standards Board, Guidance on the use of AI and other technologies (May 2026), Core Duty 7. Law Society of Ontario, Generative AI: your professional obligations (2024), practice tip 2. New Zealand Law Society, Lawyers and generative AI (March 2024).
Status. A proposed guideline, put forward by Greg Fourie SC for discussion. It has not been adopted by the Legal Practice Council, the General Council of the Bar or any society of advocates. Read the Code and the judgments themselves before relying on them.