AI Chatbot for Law Firms: Intake Use and Ethics Risk

An AI chatbot for law firms is software that answers the website chat window, asks screening questions, and hands the visitor to a form, the phone or the calendar.

It is one of three ways to staff chat; the live chat for law firms guide covers the widget and staffing generally, and this page asks the narrower question: can software handle intake, and where does the ethics risk sit.

What the numbers say about law firm chatbots

In Clio's 2024 Legal Trends Report, as summarized by 2Civility, the Illinois Supreme Court Commission on Professionalism, only 7% of the 500 US firms contacted used website chatbots.

Attitudes run ahead of adoption: 51% of prospective clients agreed chatbots can be a helpful starting point, and three in five would use one only if they could switch to a human.

The leak a chatbot answers is real: in that test, 40% of firms answered the phone and only 33% replied to email, down from 56% and 40% in 2019.

Firm-side AI use is broad but generic: Clio's 2025 Legal Trends Report found 79% of legal professionals use AI in their firms, while only 40% use a legal-specific AI tool, down from 58% in 2024.

Prospective clients bring their own AI: in Clio's 2025 consumer survey of 1,000 consumers, most had used or would consider using AI for a legal question, and of those who did, about half got their question answered, 28% were directed to contact a lawyer, and in 12% of cases an AI convinced consumers their problems weren't worth pursuing (Clio, 2025).

Chatbot, human chat or AI receptionist

The staffing choice decides what chat can do, and the phone-side version of this trade-off is the AI receptionist for law firms guide.

Both sides point at one design rule: the bot's job is capture and routing, with a visible path to a person.

Can an AI intake chatbot handle law firm intake

The capture half is mechanical: name, contact details, matter type and a booked time; the handoff lands in law firm intake, where fit, conflicts and the consultation get decided.

The rules reach the rest, starting with ABA Model Rule 1.18: a person who consults a lawyer about possibly forming a client-lawyer relationship is a prospective client, and what was learned may not be used or revealed even when no relationship follows, except as Rule 1.9 allows for a former client.

If what the firm learns could be significantly harmful, 1.18(c) bars representing an adverse party in the same or a substantially related matter, with the conflict imputed to the firm; the state's adopted version controls, and California writes its own.

The trigger is Comment [2]: a consultation is likely to have occurred if a lawyer, through advertising in any medium, specifically invites submission of information about a potential representation without clear warnings that limit the lawyer's obligations, and a person responds (comment text as written in Minnesota's Rule 1.18, which follows the ABA's).

Many firms therefore put a no-attorney-client-relationship notice beside chat and forms; the comment describes the mechanic, not a guaranteed safe harbor, and state versions vary.

ABA Formal Opinion 10-457 (August 2010) reads websites the same way: a site that specifically invites submissions about a possible representation creates a Rule 1.18 discussion when the visitor submits, a site listing only contact information does not, and site features can invite, encourage, limit or discourage the flow.

The opinion predates chatbots, so applying it to one is an analogy, not a ruling.

ABA Formal Opinion 512 (July 29, 2024) adds the AI layer: a client's informed consent is required before information about a representation enters a self-learning generative AI tool, and generic engagement-letter language is not sufficient.

Our practical point, not a rule: let the bot collect only what a fit and conflicts check needs before it asks for case details.

ChatGPT-style chatbots on a law firm website

When a general-purpose model sits behind the chat window, the firm's existing duties still apply; Clio's 2025 finding that only 40% of legal professionals use a legal-specific AI tool measures firms' own AI use, not what runs website chat.

These are the same website-wide rules the law firm website ethics guide collects, and ABA Formal Opinion 512 applies them to generative AI, naming duties of competence, confidentiality, communication, supervision, meritorious claims, candor to the tribunal and reasonable fees.

Supervision reaches the vendor: ABA Model Rule 5.3 covers nonlawyers employed or retained by or associated with a lawyer, requires measures giving reasonable assurance their conduct fits the lawyer's obligations, and makes a lawyer responsible for violations the lawyer orders, ratifies, or knows of and fails to fix in time (ABA CPR chart, as of February 2022; Formal Opinion 501 applies the rule to marketing firms and lead generators).

Opinion 512 also notes that issues under Model Rule 7.1, on communications about a lawyer's services, may surface with generative AI chatbots, citing Florida Bar Ethics Opinion 24-1, which it describes as discussing GAI chatbots under Florida Rule 4-7.13, on misleading content and unduly manipulative or intrusive advertisements; Florida is the cited example, not a template, and your state's rule controls.

On solicitation, Comment [2] to ABA Model Rule 7.3 (August 2018 text) places chat rooms, text messages and other written communications that recipients can easily disregard outside live person-to-person contact, and the solicitation rule guide covers the rest of 7.3.

That placement makes no message lawful: the rest of Rule 7.3 still applies, and states write their own versions.

Where to put a chatbot and what to measure

Nielsen Norman Group found that people dislike pop-ups and advises never showing one before visitors get value from the page (NN/g, 2019); a self-opening chat window is that pattern in chat, our observation, not NN/g's, and a corner badge the visitor starts is the least intrusive version.

Illustrative: say a month brings 40 chats that become 12 contact records and 6 consultations: the last number pays for the tool, it comes from your intake log, and no independent law firm chat benchmark was found.

When chat placement runs as a CRO test here, it is judged by consultations booked and cases signed from the firm's intake or CRM numbers; More Booked Cases tests the website side and does not staff, answer or train chat.

These rules are described, not applied: confirm how your state's versions treat a chatbot's intake flow with your state bar's ethics counsel or an ethics lawyer.

Frequently asked questions