ChamberOne

Generative AI guidance for lawyers: how six markets compare

By the ChamberOne team 5 minute read

A managing partner with clients in Colombo, London and Singapore does not want six different AI policies. She wants one policy that satisfies the strictest expectation. To write it she needs to know what each market has said, and the honest answer is that this changes often and varies in form. This article does not summarise each rulebook. It shows where to look, how to compare what you find, and which themes appear almost everywhere.

Key points

  • Any printed comparison of AI guidance goes stale quickly, so build a process for finding and recording it.
  • Check four sources in each market: the regulator, the bar or law society, the courts and the data protection authority.
  • Record the status and audience of each document, because guidance and binding direction carry different weight.
  • Competence, confidentiality, accuracy, supervision, client communication and fair fees recur across markets.
  • Write one baseline policy to the strictest expectation and add local notes where needed.

Why a country-by-country table goes stale

Guidance on AI is being issued, updated and withdrawn at a pace that makes any printed comparison unreliable within months. Regulators publish statements, courts issue practice directions, bar bodies post opinions and law societies circulate notes to members. Some of this is binding. Much is not.

A firm that treats all of it as binding will over-restrict itself, and a firm that treats all of it as optional will be caught out. Read each document for what it says about its own status.

A table also hides differences in weight. A court direction on citations has a different force from a regulator's risk report or a professional body's discussion paper. When you compare markets, record the type of document as well as its content.

Four kinds of source to look for in each market

For every country where you practise or advise, check each of the following, and note the date you looked.

  • The regulator of the profession, which may publish risk reports, warning notices or guidance.
  • The bar or law society, which often issues opinions, toolkits or ethics notes.
  • The courts, which may issue practice directions or notices about AI in filings.
  • The data protection or privacy authority, which may have views on personal data entering AI systems.

What the UK shows about the type of material

England and Wales offers a useful illustration of how these sources differ. The Solicitors Regulation Authority published a Risk Outlook report on the use of AI in the legal market. Among other things, it says firms remain responsible for AI outputs, and that firms cannot delegate accountability to an IT team or external provider. It lists issues for firms to consider, including testing, staff training and rules, verification and data protection.

The judiciary has separately published guidance for judicial office holders. It tells judges not to put private information into public AI tools, warns of inaccurate or hallucinated output and stresses personal responsibility for material produced in their name. Its latest version replaced an earlier one, which shows how quickly this material is revised.

Two documents, from two bodies, with different audiences. That pattern, of a regulator speaking to practitioners and the courts speaking to judges, will appear in other markets in different forms.

Notice also what these documents do not do. Neither is a rulebook for what your firm must do in every case, and neither replaces your own duty to read the professional rules that apply to you. They are signals of what regulators and courts expect to see, which is useful when you set your own standards.

SRA Risk Outlook report
  • Speaks to firms and practitioners.
  • Says firms remain responsible for AI outputs.
  • Discusses opportunities and risks of adopting AI.
Judicial guidance
  • Addressed to judges and those who support them.
  • Says not to put private information into public AI tools.
  • Warns of inaccurate or hallucinated output.
Two documents from England and Wales, two audiences

How to compare what you find

Build a simple grid with one row per market and the same columns each time. Fill it from the primary documents, not from commentary.

  • Who issued it and what is its status: binding, guidance or discussion?
  • Who is it addressed to: lawyers, firms, judges or the public?
  • What does it say about confidentiality and client data?
  • What does it say about checking output and citing authority?
  • Does it address supervision, training or disclosure to clients or the court?
  • When was it published or last updated?

The duties that keep appearing

Without asserting what any one rulebook requires, a handful of themes recur in discussions of AI and legal practice in many places. They are a sensible backbone for a cross-border policy.

  • Competence: understand a tool well enough to use it responsibly and to know its limits.
  • Confidentiality: know where client information goes before it goes into a tool.
  • Accuracy and candour: verify what you put before a court, a client or the other side.
  • Supervision: remain responsible for the work of staff and of tools, and keep review in place.
  • Communication: consider what clients should be told about how their matter is handled.
  • Fees: be fair and open when efficiency changes the cost of work.

Writing one policy for several markets

Start from the strictest expectation you find on each theme and apply it everywhere, then add market-specific notes where a local rule or court direction requires something more. A single baseline is easier to train people on and easier to defend.

Suppose a firm with offices in three of the six markets finds that one of them expects more disclosure to clients than the others. Applying that standard everywhere costs a few lines in an engagement letter and removes a source of confusion for lawyers who work across offices.

Name a person in each market who watches for new material and tells the others. Review the policy at set intervals and whenever a regulator or court in one of your markets changes its position.

A shared page with the date each source was last checked, and a short note on what changed, is usually enough. The point is that a change in one market reaches the others within days, not at the next annual review.

  1. 1A named person in each market watches for new material
  2. 2Tell the other markets
  3. 3Update the shared page with the date checked
  4. 4Note what changed
  5. 5Review the policy at set intervals
  6. Then back to step 1.
Keeping one policy current across markets

Local advice for local rules

A cross-border firm should not rely on a general article, this one included, for what the rules require in a particular country. Go to the regulator, the bar and the courts in each place, and take local advice where the stakes justify it.

For Sri Lanka, India, Singapore, Australia and the United States, check the guidance from the relevant regulator or bar, and for courts, the practice directions of the court in which you appear.

Sources

  1. SRA: Risk Outlook report, The use of artificial intelligence in the legal market
  2. Judiciary of England and Wales: Artificial Intelligence (AI) judicial guidance

This article is general information about the law, legal practice and technology, current when it was published. It is not legal advice. Laws and guidance change: check the sources linked above and take advice on your own circumstances from a lawyer qualified where you are.

Read next

Generative AI guidance for lawyers across six markets