Key takeaways
- Opinion 24-1 lets Florida lawyers use generative AI if they protect confidentiality, provide accurate and competent services, avoid improper billing and follow advertising rules[1].
- A chatbot that talks to clients or third parties must say it is an AI program and not a lawyer or employee of the firm[2].
- Treat the chatbot like nonlawyer intake staff: identify its status, ask factual questions only, give no legal advice and pass legal questions to a lawyer[3].
- Screen early for people who already have a lawyer, and design the flow so a prospective-client relationship never forms by accident[4][5].
What the opinion covers
The Florida Bar issued Ethics Opinion 24-1 on January 19, 2024. It concludes that lawyers may use generative AI in practice, provided they protect client confidentiality, provide accurate and competent services, avoid improper billing practices and comply with the restrictions on lawyer advertising[1]. The opinion speaks to research and drafting tools, but its most specific guidance concerns tools that talk to the public: website chatbots and other client-facing AI.
Florida matters well beyond its borders. It had 17,261 offices of lawyers in 2023, second only to California[6], and its opinion is one of the most cited by other bars considering intake tools.
The disclaimer requirement
Opinion 24-1 says generative AI chatbots that communicate with clients or third parties must comply with the lawyer advertising rules and must include a disclaimer indicating that the chatbot is an AI program and not a lawyer or employee of the law firm[2]. That is a requirement, not a suggestion, and it has two parts: the bot is AI, and it is not a lawyer or staff member.
Opening message for a chat widget: βHi, I'm the Seaside Law Group virtual assistant, an AI program. I'm not a lawyer or a member of the firm's staff, and I can't give legal advice. I can take some details and arrange for our team to contact you.β Seaside Law Group is fictional; use your own firm name.
Public expectations point the same way. 76% of Americans say it is extremely or very important to be able to tell whether content was made by AI or by people[7], yet 53% are not confident they can tell[8]. A clear disclaimer settles a question most visitors cannot answer for themselves.
Source: Pew Research Center, 2025[7]
Show the data
| Item | Value |
|---|---|
| Say it is extremely or very important to be able to tell | 76% |
| Are not too or not at all confident they can tell | 53% |
Treat the bot like nonlawyer intake staff
The opinion points to Florida Ethics Opinion 88-6 as especially useful for intake chatbots. Under 88-6, nonlawyers may conduct an initial interview with a prospective client, but they must clearly identify their nonlawyer status, limit questions to obtaining factual information, not offer legal advice and refer legal questions to the lawyer[3]. Applied to a chatbot, those four limits become design requirements.
| 88-6 limit | What the chatbot does | What it must not do |
|---|---|---|
| Identify status | Opens with the AI and not-a-lawyer disclaimer | Use a human name or photo that implies a person |
| Factual questions only | Collects contact details, matter type, dates, parties for conflicts | Ask the visitor to narrate every detail of the case |
| No legal advice | Explains what happens next at the firm | Assess the strength of a claim or suggest a course of action |
| Refer legal questions | Offers a consultation request or callback | Answer βdo I have a case?β or βwhat should I do?β |
Avoid accidental relationships
The opinion warns that intake chatbots risk creating a prospective-client relationship, or even a lawyer-client relationship, without the lawyer knowing[4]. A bot that collects a detailed story creates confidentiality duties to that person and can create conflicts that disqualify the firm from representing someone on the other side. A bot that responds with anything resembling advice invites the visitor to rely on it.
- Collect only what the firm needs to run a conflict check and decide whether to call back.
- Run the conflict check before anyone asks for the full story.
- End every chat with a plain statement that the firm has not agreed to represent the visitor.
- Send the transcript to a named person at the firm, not to a shared inbox nobody reads.
Screen for people who already have a lawyer
Opinion 24-1 says a lawyer should consider screening questions that limit the chatbot's communications if the person is already represented by another lawyer[5]. The concern is contact with represented persons and interference with an existing relationship. Put the question early: βDo you already have a lawyer for this matter?β If the answer is yes, the bot should explain that the firm can speak with them about a second opinion or a change of counsel only through the proper channels, take contact details and stop gathering facts.
Advertising, confidentiality and billing
Because a client-facing chatbot is a communication about the firm's services, its scripts are subject to Florida's advertising rules. Claims about results, specialization or how quickly the firm responds must be accurate and permitted, and the bot must not make statements the firm could not put in an advertisement.
Confidentiality runs through the whole design: the firm needs to know where chat transcripts are stored, who at the vendor can see them and whether they are used to train models. On billing, the opinion says generative AI does not permit improper practices such as double-billing[9]; time a lawyer spends reviewing chatbot transcripts is real work, but time the bot saved is not billable.
The same rules apply to voice
Florida's opinion speaks of chatbots, but a firm should apply the same reasoning to any AI that converses with the public, including AI that answers the phone. The disclosure becomes part of the greeting, the questions are limited to facts, legal questions are routed to a lawyer and the call ends with a clear next step. If an SI Employee such as Ava answers your Florida line, the greeting your firm approves says that Ava is the firm's AI assistant, not a lawyer, and that the call is recorded; Florida requires every party's consent to record a call.
Testing before launch
Script review
A lawyer approves the opening disclaimer, every question and every fixed response.
Adversarial tests
Staff try to get the bot to give advice, predict outcomes, discuss fees it should not quote or keep chatting with a represented person.
Conflict check timing
Confirm the bot collects names for conflicts before inviting any narrative.
Transcript routing
Check that every conversation reaches a named person within a set time.
Monthly review
Read a sample of real transcripts and adjust the script where it drifted.
Questions to ask a chatbot vendor
Opinion 24-1's confidentiality and competence duties mean the firm must understand the tool well enough to supervise it. Get written answers before signing:
- Where are transcripts stored, for how long, and can the firm delete them on request?
- Are conversations used to train models that serve other customers?
- Which vendor staff can read transcripts, and for what purposes?
- Can the firm fix the opening disclaimer so it always appears, whatever the model generates?
- Can the firm restrict topics, block outcome predictions and force a handoff when legal questions appear?
- How does the tool behave when it is unsure: does it say so, or does it guess?
A vendor that cannot answer these questions in writing is asking the firm to supervise something it cannot see.
What to log and keep
Supervision needs evidence. Keep the approved script with its version date, the test results from before launch and each later change, and a record of the transcript reviews. When a visitor later claims the bot told them something, the firm can show what the bot was configured to say at the time and how the conversation was handled.
Retention should match how the firm treats other intake records. Transcripts of declined inquiries hold confidential information from prospective clients, so they need the same access limits and deletion schedule as intake notes and call recordings.
Example: a personal injury firm's website chat
Example: a fictional Orlando personal injury firm launches website chat. Its first version greets visitors with a friendly first name and photo, and in testing it tells a visitor that a two-year-old fall is probably still within the time limit. The firm pulls it. The second version opens with the AI disclaimer, asks whether the visitor already has a lawyer, collects contact details, the date of the incident and the other party's name for a conflict check, and answers any legal question with an offer of a callback from an attorney. Every transcript goes to the intake manager, who reviews a sample weekly.
For how other states compare, see California's generative AI guidance and Texas Opinion 705.
Frequently asked questions
What does Florida Bar Opinion 24-1 require for chatbots?
A chatbot that communicates with clients or third parties must comply with advertising rules and include a disclaimer that it is an AI program and not a lawyer or employee of the firm.
Can a Florida law firm chatbot answer legal questions?
It should not. Following Opinion 88-6, intake by nonlawyers is limited to factual questions, with legal questions referred to a lawyer. Design the bot to offer a consultation or callback instead.
How can a chatbot create an attorney-client relationship?
By collecting detailed facts or appearing to give advice, so the visitor reasonably believes the firm is helping them. Limit questions, run conflicts first and state that the firm has not agreed to represent the visitor.
Does Opinion 24-1 apply outside Florida?
It binds no other state, but many bars and firms use it as a model, and its intake limits match general rules on prospective clients and nonlawyer assistance.
References
The Florida Bar (2024). Florida Bar Ethics Opinion 24-1 (opens in a new tab) β
The Florida Bar (2024). Florida Bar Ethics Opinion 24-1 (opens in a new tab) β
The Florida Bar (2024). Florida Bar Ethics Opinion 24-1 (opens in a new tab) β
The Florida Bar (2024). Florida Bar Ethics Opinion 24-1 (opens in a new tab) β
The Florida Bar (2024). Florida Bar Ethics Opinion 24-1 (opens in a new tab) β
U.S. Census Bureau (2023). County Business Patterns 2023: state file (cbp23st) (opens in a new tab) β
Pew Research Center (2025). How Americans View AI and Its Impact on Human Abilities, Society (opens in a new tab) β
Pew Research Center (2025). How Americans View AI and Its Impact on Human Abilities, Society (opens in a new tab) β
The Florida Bar (2024). Florida Bar Ethics Opinion 24-1 (opens in a new tab) β