Judgment, not just content: what ethical AI use looks like at a South African law firm

Editorial illustration: AI-generated; not an actual SD Law staff or client photograph.

Ethical AI use in a South African law firm requires accuracy, confidentiality, fairness and accountable legal judgment. AI may help with research and drafting, but a responsible lawyer must verify sources, assess the facts and stand behind the advice. Neither speed nor improved quality is guaranteed.

Most discussion of artificial intelligence in law asks three questions: is it accurate, is it safe, is it biased? These questions are essential to client protection, but they do not exhaust the issue. Clients also need to know whether the technology improves the advice they receive. To answer that question, we must look beyond document production and examine how AI affects the lawyer’s reasoning.

Efficiency depends on the task

Lawyers may use AI to assist with transcript summaries, research questions, document organisation or preparation for consultations. Whether this saves time depends on the task, the tool and the verification required. In a complex matter, checking and correcting the output may reduce or eliminate the initial saving.

Consider a hypothetical custody matter with an international dimension. A child’s welfare is at stake and there is no room for a plausible but wrong answer. AI may help organise an anonymised chronology, but confidential records and sensitive personal information must not be entered into unapproved tools. Appropriate privacy, security and professional safeguards come first. Any permitted AI-assisted summary remains a starting point: every material claim must be checked against the original record.

The benefit may be a more considered analysis, provided the lawyer interrogates the output and checks the underlying evidence. That improvement is possible, not automatic. AI does not relieve a lawyer of intellectual or professional responsibility. It introduces another input that needs judgment.

Where the real value lies: judgment, not speed

AI can expose alternative ways of organising an argument or identifying questions worth researching. That breadth can be useful, but generated patterns are not verified precedent or a complete statement of the law. The lawyer must consult authoritative sources and determine what fits this client, this matter and these facts.

AI cannot replace a lawyer’s judgment. Equally, competent legal advice does not depend on using AI: careful research, experience and sound reasoning remain essential with or without it. Where a tool is appropriate, its suggestions must be tested against the evidence and current law before the lawyer adopts them.

What this looks like in practice

The principle is simple, but it must be applied through standards that can be explained. An AI-assisted draft requires substantive review and rewriting where necessary, not merely a change of tone. Material facts must be checked against the file, citations against the primary source, and recommendations against the client’s circumstances and objectives. The responsible lawyer must be able to explain the final advice.

If you already have an AI-written business agreement, our guide to checking an AI-drafted contract before signing explains the practical questions to resolve.

Using AI as a tool does not make the ethical questions disappear. Errors, fabricated citations, outdated law, bias and exposure of confidential information remain risks. The lawyer must assess those risks before use and must not treat fluent output as authoritative legal judgment.

What clients should expect from a firm that uses AI ethically

Clients should expect concrete safeguards, not a promise that the technology itself is trustworthy:

• Confidentiality is preserved. The firm should assess confidentiality, data protection and security before use. Confidential or sensitive information must not be placed in unapproved tools.
• A named lawyer is accountable for every document that leaves the firm. A human is responsible and a system is not blamed for any mistakes.
• AI is treated as an input to judgment, never a substitute for it. Nothing goes to a client unread, unchecked and unowned.
• The process is honestly described. The firm should explain the role of AI honestly. Time, cost and quality benefits vary, and professional review remains essential. Clients deserve to know what they are getting.

Trust is earned by the process, not the tool

South African lawyers remain subject to the Legal Practice Council Code of Conduct, including duties of competence, confidentiality and proper control and supervision. AI does not displace those obligations. The tool may be new; responsibility for the work remains with the legal practitioner.

Clients need advice they can understand and rely on, not merely access to a tool. They engage a lawyer to assess evidence, identify legal and practical risks, explain options and take responsibility for the recommendation. Our related commentary on training lawyers to exercise judgment considers how firms can develop that capacity.

SD Law can help

At SD Law, we welcome tools that enhance our service to you, our clients. We embrace cutting-edge technology as part of our vision to be a modern, client-driven law firm, and we do so responsibly and ethically. Personal contact with our clients remains a core value of our practice, and we will always be driven by judgment. If you have a legal matter you’d like to discuss, contact Simon on 086 099 5146 or email sdippenaar@sdlaw.co.za for a confidential, personal discussion.

Simon Dippenaar

Founder & Managing Partner, SD Law

Attorney Simon Dippenaar has a BBusSc LLB degree and Professional Diploma in Legal Practice from the University of Cape Town, and is an admitted attorney of the High Court of South Africa. He is the founder and director of private legal practice, Simon Dippenaar & Associates, with offices in Cape Town and Gauteng representing South African and international clients.

Read articles