Where the Law Society of Ireland Stands on AI

Solicitors considering AI often assume there is a rule somewhere telling them not to. There is not. The position taken by the Law Society of Ireland has been consistent and unsurprising: generative AI has legitimate uses in legal practice, and those uses sit inside your existing professional obligations rather than outside them.

That is a more useful position than a prohibition would have been, and a more demanding one than a blanket approval.

The consistent message

Across the Society’s published material and the coverage in the Law Society Gazette, three themes recur:

Supervision

AI output is not a work product until a qualified person has reviewed it. The technology prepares; the solicitor remains responsible. Nothing about adopting AI transfers professional responsibility to a vendor.

Confidentiality

Client information must be protected to the same standard regardless of the tool involved. The obligation does not soften because the software is new or convenient.

Professional judgment

Advice remains the solicitor’s. AI can assemble, summarise, draft and structure. It does not exercise judgment, does not hold a practising certificate, and does not carry a duty to the court or the client.

What the guidance does not say

It is worth being clear about the absences, because they are where the misconceptions live.

• It does not prohibit generative AI in legal practice.
• It does not require client consent as a blanket precondition for every use.
• It does not endorse or approve specific products. Evaluating a given tool remains your responsibility.
• It does not create a separate AI compliance regime. Your existing duties are the framework.

What this means on a Tuesday morning

Regulatory guidance is only useful if it translates into behaviour. In practice, a solicitor acting consistently with the Society’s direction would:

• Use business-tier tools with appropriate data terms rather than free consumer accounts for client-related work
• Review every AI-assisted output before it leaves the practice, and be able to say who did
• Keep a short written note of how AI is used in the practice and by whom
• Train staff on what may and may not be entered into an AI tool
• Treat AI output as a first draft from a capable but unqualified assistant — useful, fast, and always checked

None of that is burdensome. Most of it is what a well-run practice does with any new process.

Writing down your position

One practical step does more than any other to demonstrate careful adoption: put your position in writing. Not a policy document that takes a month to draft — a single page that answers four questions.

• Which tools are approved for use in the practice, and on which tiers
• What may and may not be entered into them — particularly around identifiable client information
• Who reviews AI-assisted output before it leaves the practice
• Who to ask when someone is unsure

The value is partly in the document and largely in the act of writing it. A practice that has thought these questions through behaves differently from one that has not, and the difference shows up in the decisions people make when nobody is watching.

It also answers the question a client, an insurer or the Society might reasonably ask: how do you supervise this? Being able to hand over a page beats reconstructing your reasoning after the fact.

The risk of doing nothing

There is a quiet assumption that inaction is the conservative choice. It is not obviously true. A practice that ignores AI entirely is not avoiding risk; it is accepting a different one — slower turnaround, higher admin cost per matter, and eventual competition from practices whose cost base has fallen.

The careful path is not avoidance. It is deliberate adoption with the controls in place.

The short version

The Law Society’s position supports careful use. Supervision, confidentiality and professional judgment are the conditions. Meet them, and you are working within the profession’s expectations rather than at the edge of them.