Disclosing AI Use to Clients: What to Say and When

Clients are starting to ask whether their lawyers use AI, and some are setting rules about it. A general counsel may want to know which tools touch the company's data. An individual client may worry that their matter is being handled by software. Disclosing AI use to clients thoughtfully, before they have to ask, can prevent misunderstandings and help you meet duties around communication, confidentiality and fees.

This guide covers when disclosure may be required or advisable, what to put in engagement letters, how to talk about it in conversation and how to handle client restrictions. It is written for lawyers and firm leaders deciding how to approach the subject.

This article is general information, not legal advice. Disclosure expectations vary by jurisdiction and are still developing.

When disclosure may be required

There is no single rule that says "always disclose" or "never disclose." Bar guidance tends to tie the question to existing duties.

ABA Model Rule 1.4 requires lawyers to reasonably consult with clients about the means used to achieve their objectives and to explain matters enough for clients to make informed decisions. ABA Formal Opinion 512 discusses several situations where telling clients about generative AI use may be necessary, including:

• When confidential client information will be entered into a tool, which may call for informed consent depending on the tool and circumstances.
• When the client asks directly how work is being done.
• When AI use affects the basis or amount of fees.
• When the use is significant to the representation, for example if AI output materially informs strategy or advice.

State bars have issued their own opinions with differing emphasis. Check yours.

When disclosure is usually advisable even if not required

• The client has outside counsel guidelines that mention AI.
• The matter involves highly sensitive personal, medical or financial information.
• You plan to use AI in ways the client might not expect, such as analysing their documents at scale.
• The client is less familiar with legal practice and may misunderstand how work is produced.

What to put in the engagement letter

A short paragraph in your standard engagement letter sets expectations early. It might cover:

1. That the firm may use technology tools, including AI, to assist with tasks such as research, document review and drafting.
2. That lawyers review and remain responsible for all work.
3. How client information is protected, such as using only tools the firm has vetted for confidentiality.
4. How AI-assisted work affects billing.
5. That the client can ask questions or request that AI not be used on their matter.

Keep the language plain and specific. A vague line that "the firm uses various technologies" does little to inform anyone.

Talking about it in conversation

Clients often have three underlying questions:

Client question	What they usually mean	A useful answer covers
"Are you using AI on my matter?"	Is a lawyer still in charge?	Which tasks AI helps with and that a lawyer reviews everything
"Is my information safe?"	Where does my data go?	Which tools, what protections, whether data is used for training
"Will this make it cheaper?"	Am I paying for machine time?	How billing reflects AI-assisted work

Be honest about limits as well as benefits. Telling a client that you verify every citation because AI tools can make errors tends to build confidence, not reduce it.

Handling client restrictions

Corporate clients may restrict AI in several ways:

• Prohibitions on certain tools or categories of tools.
• Notice requirements before AI is used on their matters.
• Approval requirements for specific tools.
• Data rules, such as no client data in tools hosted in certain countries.

Make sure those requirements are recorded in the matter file and communicated to everyone working on it. A restriction that lives only in a partner's inbox is easily breached.

Billing transparency

AI changes how long some tasks take. Clients may reasonably ask whether a faster first draft means a lower bill. Formal Opinion 512 indicates that lawyers billing hourly should bill for actual time spent and generally should not bill clients for learning to use tools they will use across their practice. Some firms are moving to flat fees for tasks where AI changes the time profile. Whatever approach you take, explain it upfront.

Choosing tools that are easier to explain

Disclosure conversations are simpler when you can describe exactly how a tool handles data. eLaw is designed for law firms and legal teams and describes its approach as cited answers with firm-isolated data. Being able to say that clearly, and back it up with contract terms, makes client conversations more concrete.

Draft your disclosure language

1. Review your engagement letter template and add a plain AI paragraph.
2. Prepare short answers to the three common client questions.
3. Check active matters for client guidelines that mention AI.
4. Decide who handles requests to opt out of AI use.

FAQ

Do lawyers have to tell clients they use AI?

Not in every situation. Guidance generally links disclosure to existing duties, such as communication, confidentiality and fees. Disclosure or consent may be needed when confidential information goes into a tool, when the client asks, or when AI affects fees or advice. Rules vary by jurisdiction, so check local opinions.

Can a client refuse to let their lawyer use AI?

Clients can set conditions on how their matter is handled, and many corporate clients do so in outside counsel guidelines. If a client asks that AI not be used, the firm should record the instruction and make sure everyone on the matter follows it, or discuss alternatives if that is impractical.

Should AI use be mentioned on invoices?

Some firms describe AI-assisted tasks in invoice entries for transparency; others address it in the engagement letter only. The key is that billing is honest and consistent with what the client agreed. Describing tasks accurately helps avoid disputes about time charged for work that AI sped up.

What if a client discovers AI use they were not told about?

Explain promptly and plainly which tasks were involved, how information was protected and how the work was reviewed. Check whether any client guidelines or consent requirements were breached. Updating engagement terms and internal processes can help prevent a repeat.