Key takeaways
- People want to know: 76% of Americans say it is extremely or very important to tell whether content was made by AI or by people[1].
- Florida requires AI chatbots that talk to clients or third parties to say they are an AI program, not a lawyer or employee[2].
- Disclose in the first sentence, again before collecting details, and every time someone asks.
- Outbound AI-voice calls fall under TCPA rules on artificial voices, so consent rules apply[3].
Why disclosure matters to callers
Most people cannot tell whether they are dealing with AI, and most want to. Pew found 76% of Americans say it is extremely or very important to be able to tell whether pictures, videos and text were made by AI or by people[1], 53% are not too or not at all confident they could tell[4], and about six in ten would like more control over how AI is used in their own lives[5]. A caller who discovers halfway through that they have been talking to a machine is likely to feel misled, even if nothing false was said.
For a law firm, the stakes are higher. A prospective client may share sensitive facts because they believe a person at the firm is listening. Clear disclosure protects the caller's choice and the firm's credibility.
What the rules say
- Florida Bar Opinion 24-1 says generative AI chatbots that communicate with clients or third parties must include a disclaimer that the chatbot is an AI program and not a lawyer or employee of the firm[2].
- Florida Ethics Opinion 88-6, which Opinion 24-1 applies to intake chatbots, says nonlawyers doing initial intake must identify their nonlawyer status[6].
- The FCC ruled in February 2024 that calls made with AI-generated voices are "artificial" under the Telephone Consumer Protection Act[3], and telemarketers need prior express written consent before robocalling consumers, a standard the ruling applies to AI voices[7].
- Some states have their own laws on disclosing bots in commercial or political communications. Check your state and the states your callers are in.
Even where no rule requires it, disclosure is the safer default for any firm. It costs one sentence.
Where to place the disclosure
- In the first sentence of every call or chat, before any question.
- Again before collecting sensitive details, in a short reminder.
- Immediately and truthfully whenever a caller asks if they are talking to a person.
- Before transferring to a human, so the caller knows the change has happened.
- In any written follow-up, such as a confirmation text or email, if the AI sent it.
- In the voicemail greeting or website text if callers will reach the AI system from there.
Sample wording by moment
| Moment | Sample wording | Avoid |
|---|---|---|
| Call opening | "You've reached the firm. I'm an automated AI assistant, not a lawyer, and I can't give legal advice." | Opening with a human name and no disclosure |
| Chat opening | "Hi, I'm the firm's AI assistant. I'm not a lawyer or staff member. I can take details so the team can follow up." | A photo of a person beside the chat window |
| Before sensitive details | "As a reminder, I'm an AI assistant. Please share only what you're comfortable with; a lawyer will go over details later." | Asking for the full story with no reminder |
| Asked "Are you real?" | "I'm an AI assistant, not a person. Would you like me to take your details for a call back from the team?" | Deflecting or joking |
| Before transfer | "I'm going to connect you with a member of our team now." | A silent hand-off |
| Follow-up text | "Sent by the firm's automated assistant. Reply STOP to opt out." | Signing as a named staff member |
Write it so people actually hear it
- Keep it short. One clear sentence beats a paragraph nobody listens to.
- Use plain words: "automated AI assistant", "not a lawyer". Avoid jargon such as "virtual agent".
- Say it before the first question, not after.
- Speak at a normal pace. Disclosures read quickly sound like small print.
- Do not undercut it with chatty filler that makes the system sound human.
Can the assistant have a name?
A name helps callers refer to the assistant and makes transcripts easier to read, but it must not imply a person. Pair the name with the disclosure every time, such as "I'm the firm's AI assistant", and avoid surnames, job titles or backstories that suggest a staff member. If your firm uses a branded product name, use it as the name of the assistant, not as a person's name.
Handling "Am I talking to a real person?"
Sample response: "No, I'm an AI assistant for the firm, not a person. I can take your details so someone from the team can call you back, or you can leave a voicemail for staff. Which would you prefer?"
Answer immediately and give the caller a choice. Some callers will continue happily; others want to leave a message for a person. Both are fine. What damages trust is a system that dodges the question.
Inbound calls versus outbound calls
Most law firm AI phone use is inbound: the caller rang the firm. Disclosure there is about honesty and professional rules. Outbound calls are different. Because the FCC treats AI-generated voices as artificial under the TCPA[3], automated outbound calls to consumers trigger consent requirements, and marketing calls need prior express written consent[7]. If you plan any outbound AI calls, such as callbacks or reminders, get advice on consent and keep records of it before the first call is made.
Test your disclosure
Call it yourself
Listen to the first 15 seconds as a stranger would. Is it obvious you are talking to AI?
Ask the question
Ask "Am I talking to a real person?" at three different points in the call.
Try a chat
Open the website chat on a phone and check the disclosure is visible without scrolling.
Check the paperwork
Read the follow-up texts and emails and confirm none are signed as a person.
Review transcripts
Each month, read a sample of transcripts for any moment where the disclosure was skipped or blurred.
Website, voicemail and chat
Disclosure should be consistent wherever people meet the system. If your website says "Call us any time", add that out-of-hours calls are answered by the firm's AI assistant. If your voicemail greeting routes callers to the assistant, say so in the greeting. On chat, show the disclosure in the first message and keep a short label visible above the input box, so a visitor who scrolls past the greeting still sees it.
Disclosure in other languages
If the assistant speaks Spanish or another language, write and approve the disclosure in that language too, rather than relying on automatic translation of the English version. Have a fluent speaker check that "not a lawyer" and "can't give legal advice" come across clearly, because a loose translation can turn a disclaimer into something that sounds like a job title.
Pair it with the recording notice
Many firms record AI-answered calls, and recording consent has its own rules that vary by state. Combine the two notices in the opening so callers hear both before any question: who they are speaking with, and that the call is recorded. Keep the combined opening under about 20 seconds, and have a lawyer approve it once for all channels so staff are not improvising variations.
Who signs off
Disclosure wording is client communication, so a lawyer should approve it, just as a lawyer approves intake scripts and advertising. Record the approved wording, the date and the approver, and require the same approval for any change. When a vendor updates its product, check that the approved disclosure still plays first and has not been replaced by a default greeting. A quarterly listen to a handful of real calls is usually enough to catch drift.
Example: fixing a disclosure that nobody heard
Example (fictional): a family law firm's AI assistant opens with a warm greeting, a firm slogan and then the disclosure in the third sentence. Transcript reviews show several callers later asking for "the lady I spoke to". The firm moves the disclosure to the first sentence, cuts the slogan and adds a reminder before the intake questions. In the next month's review, nobody mistakes the assistant for a person, and the share of callers who complete intake does not drop.
Common mistakes
- A disclosure that only appears in website terms of use.
- A stock photo of a person next to a chat window.
- A human-sounding name with no disclosure.
- Follow-up messages signed by a staff member who never saw the conversation.
- Removing the disclosure because callers "hang up more" without checking whether they do.
Frequently asked questions
Do law firms have to tell callers they are speaking with AI?
Florida requires AI chatbots that communicate with clients or third parties to disclose that they are an AI program, not a lawyer or employee[2]. Other states vary, but disclosure is the safer default.
Where should the AI disclosure go?
In the first sentence, again before sensitive details, whenever someone asks, before a transfer and in any written follow-up.
Can I give my AI assistant a human name?
A name is fine if it is always paired with a clear statement that the assistant is AI and not a person.
Do TCPA rules apply to AI voices?
Yes. The FCC ruled that AI-generated voices are artificial under the TCPA[3], so consent rules apply to automated outbound calls.
References
Pew Research Center (2025). How Americans View AI and Its Impact on Human Abilities, Society (opens in a new tab) โ
The Florida Bar (2024). Florida Bar Ethics Opinion 24-1 (opens in a new tab) โ
Federal Communications Commission (2024). FCC Makes AI-Generated Voices in Robocalls Illegal (opens in a new tab) โ
Pew Research Center (2025). How Americans View AI and Its Impact on Human Abilities, Society (opens in a new tab) โ
Pew Research Center (2025). How Americans View AI and Its Impact on Human Abilities, Society (opens in a new tab) โ
The Florida Bar (2024). Florida Bar Ethics Opinion 24-1 (opens in a new tab) โ
Federal Communications Commission (2024). FCC Makes AI-Generated Voices in Robocalls Illegal (opens in a new tab) โ