Compliance and Privacy10 min read
AI Intake and Prospective Client Conflicts
Your AI intake agent can turn a caller into a prospective client under Nevada Rule 1.18. Run this conflict check field audit and limiting script.
A Nevada law firm's AI intake agent can create a prospective client on the first call. Under Rule 1.18 of the Nevada Rules of Professional Conduct, as printed in the State Bar of Nevada compilation of Rules 7.1 through 7.5, 1.4 and 1.18 (April 2018), anyone who consults about possibly hiring you is a prospective client, signed or not. If the agent collects information that could be significantly harmful to that caller, the firm can be barred from the other side of the matter, so the conflict check has to run before the facts arrive.
This playbook is for managing partners and office managers running automated phone or chat intake. It covers the exposure flow, a field audit, the 1.18(d) cure, and a limiting script. It is not legal advice, and the compilation is dated April 2018, so confirm the current text before relying on it.
Why an AI intake call can trigger a prospective client conflict check
Rule 1.18(a) is short: 'A person who consults with a lawyer about the possibility of forming a client-lawyer relationship with respect to a matter is a prospective client.' It does not require the caller to reach a human. Whether a bot conversation counts as consulting 'with a lawyer' is something a bar or court would decide, so the safe assumption is that it does.
Paragraph (b) adds a duty that survives the hang-up: 'Even when no client-lawyer relationship ensues, a lawyer who has learned information from a prospective client shall not use or reveal that information, except as Rule 1.9 would permit with respect to information of a former client.' The caller who never signs still has confidentiality rights in what your bot heard.
Paragraph (c) is where the business risk sits. A lawyer subject to (b) 'shall not represent a client with interests materially adverse to those of a prospective client in the same or a substantially related matter if the lawyer received information from the prospective client that could be significantly harmful to that person in the matter.' The bar then spreads: 'no lawyer in a firm with which that lawyer is associated may knowingly undertake or continue representation in such a matter,' unless paragraph (d) is met.
A human receptionist creates the same risk. A bot just runs all night and never tires of asking follow-up questions.
The conflict exposure flow: four gates decide whether the firm is out
Run any intake call through these gates in order.
- Did the caller consult about a possible engagement? Someone describing their own problem to see whether you will take it is a prospective client, and the 1.18(b) duty starts here.
- Did the agent collect information that could be significantly harmful to the caller? Names, phone numbers, and matter type usually are not. The caller's account of events, admissions, and finances usually are. If no, the firm owes confidentiality but is not disqualified.
- Is the firm materially adverse to the caller, now or later, in the same or a substantially related matter? If no, there is nothing to cure. If yes, the receiving lawyer is disqualified, and by default so is the firm.
- Is a 1.18(d) path open? Written informed consent from both sides, the screening alternative, or the caller's up-front agreement that what they share will not bar the firm. If none applies, the firm declines or withdraws.
The rule text does not say who 'received' what a machine heard. Assume it is whoever reads the transcript.
Size changes the outcome. Assume a solo family law practitioner whose bot takes after-hours calls. A spouse calls on a Sunday, describes hidden accounts, and never retains. Two weeks later the other spouse walks in. A solo has no colleague to wall off, so the only route is written consent from both sides, which the first caller has no reason to give.
Now assume a 12 lawyer firm whose defense group already represents an insurer. A claimant calls the bot at night and narrates the accident. This firm can screen, but only if the reviewing lawyer can be walled off and the transcript never sat in an inbox the defense team reads.
Field audit: what your intake bot asks and what each answer can cost
The fastest fix is collecting less before the conflict check clears, not a longer disclaimer. The last two columns below are our recommendations, not rule text.
| Field collected | Why the bot asks | Rule 1.18 risk if the firm later takes the other side | Retain or suppress | Where it is stored |
|---|---|---|---|---|
| Caller name and phone | Callback and conflict check | Low: identity alone is rarely harmful | Retain | Conflict index and contact record |
| Names of other parties | Conflict check | Low, and it lets you decline before facts arrive | Retain | Conflict index |
| Matter type | Routing to a practice group | Low to moderate | Retain | Intake record |
| Date and place of incident | Deadlines and routing | Moderate: can place the caller in the story | Retain only if needed | Restricted intake record |
| Caller's account of what happened | Qualifying the case | High: the core of 'significantly harmful' | Suppress until cleared | Consult notes after clearance, never the bot |
| Admissions, finances, health, immigration status | Fee and case screening | High | Suppress | Nowhere in the bot platform |
| Uploaded documents or photos | Case evaluation | High | Suppress until cleared | Matter folder after clearance |
| Full audio and transcript | Quality review | Holds whatever the caller volunteered | Retain, restricted | Segregated store with access log |
The pattern is two-stage intake. Stage one takes identity and party names, runs the check, and stops. Stage two, the facts, happens after a lawyer clears the conflict. A well-built AI receptionist can run stage one at 2 a.m. and book the consult through appointment scheduling, so the caller gets speed without the firm absorbing facts it cannot unhear.
The 1.18(d) cure has two doors, plus one you open before the call
Paragraph (d) reads: 'When the lawyer has received disqualifying information as defined in paragraph (c), representation is permissible if: (1) Both the affected client and the prospective client have given informed consent, confirmed in writing, or:' and then sets out a screening alternative. Our source extract does not reproduce the screening conditions, so read them in the Nevada Legislature's posted Rules of Professional Conduct and build your wall to meet each element of the current text.
Dual consent works only when the caller has nothing at stake, which is rare. Screening works only for firms with more than one lawyer and intake records that can actually be walled.
The third option is in the same rule. A lawyer may 'condition conversations with a prospective client on the person's informed consent that no information disclosed during the consultation will prohibit the lawyer from representing a different client in the matter.' A bot can deliver this, because it happens before fact gathering. A sample limiting script:
You've reached [Firm Name]. I'm an automated assistant, not a lawyer, and talking with me does not make you a client. I'll take your name, number, and the names of the other people involved, so the firm can check for conflicts. Please don't tell me what happened yet. If you do share details, you agree that nothing you say will stop the firm from representing someone else in this matter. Do you understand and agree? Say yes to continue, or say callback and a staff member will call you.
Two cautions. Whether a recorded yes to an automated prompt is informed consent is a question for your ethics counsel, not your vendor. And a caller who declines must get the callback, with the bot collecting nothing further. Log every answer with a timestamp and the script version.
A searchable transcript can defeat the screen your cure depends on
ABA Formal Opinion 512 (July 29, 2024), on lawyers' use of generative AI, is why transcripts are not ordinary CRM data. At page 7: 'because many of today's self-learning GAI tools are designed so that their output could lead directly or indirectly to the disclosure of information relating to the representation of a client, a client's informed consent is required prior to inputting information relating to the representation into such a GAI tool.'
The sharper warning is internal. A tool 'may disclose information relating to the representation to persons in the firm (1) who either are prohibited from access to said information because of an ethical wall or (2) who could inadvertently use the information from one client to help another client, not understanding that the lawyer is revealing client confidences.'
Map that onto the 12 lawyer firm. If transcripts land in a shared inbox, a firm-wide search index, or an AI assistant that answers over all firm data, the wall leaks before anyone builds it.
The opinion also tells lawyers to vet 'the vendor's conflicts check system to screen for adversity among firm clients.' Ask your intake vendor in writing whether transcripts can be restricted per matter with an access log, and whether its AI features retrieve across all conversations. Our ABA 512 intake post covers the wider vendor questions.
Your intake script is a retained record for 3 years after last use
The same April 2018 compilation carries advertising rules that reach the bot's follow-up texts and emails. Rule 7.2(b)(2): 'All advertisements and written communications disseminated pursuant to these Rules shall identify the name of at least one lawyer responsible for their content.' A follow-up text should name that lawyer.
Rule 7.2(b)(6) makes quality statements 'subject to proof of verification, to be provided at the request of the state bar or a client or prospective client.' A bot that improvises 'we win most of these cases' has made a claim a caller can demand you prove. Lock the script so the model cannot generate quality claims.
Rule 7.2(c): 'Any statement or disclaimer required by these rules shall be made in each language used in the advertisement or writing.' A bilingual bot that switches to Spanish after an English disclaimer has a gap. Required statements, and as good practice the limiting script, need a Spanish version the bot actually reads.
Then retention. Rule 7.2A(c)(4): 'A copy of all information provided pursuant to this Rule shall be retained by the lawyer or law firm for a period of 3 years after last regular use of the information.' The conservative reading treats script versions as retained records, not disposable prompt text.
The arithmetic is manageable. Assume the 12 lawyer firm revises its prompt monthly: 12 versions a year, each kept 3 years after retirement, means 36 retired versions on file at steady state, plus the live one. A solo editing quarterly holds 4 times 3, or 12, plus the live one. Easy to store, if someone saves each version with a date, since many bot dashboards overwrite the prompt in place.
Where this analysis does not hold
If the firm never represents the other side of the matters it takes intake for, such as a plaintiff-only injury shop, 1.18(c) disqualification rarely bites, though the 1.18(b) confidentiality duty still applies. If the caller gave only identity and party names, there is likely no significantly harmful information. If the later matter is not the same or substantially related, (c) does not apply.
The text is a limit too. Our quotes come from a compilation dated April 2018, Nevada may have amended Rule 1.18 since, and the screening conditions are not quoted here. ABA Opinion 512 is persuasive, not binding in Nevada. Whether a bot conversation is a consultation 'with a lawyer' is unresolved in our sources. This is not legal advice; your ethics counsel is the authority.
Questions owners ask
Does an AI intake bot create a prospective client in Nevada?
Plan as if it does. Nevada Rule 1.18(a) covers any person who consults with a lawyer about possibly forming a client-lawyer relationship on a matter. A bot answering as your firm and inviting callers to describe their problem looks like that channel. Our sources do not settle whether a machine conversation counts, so treat every caller describing their own matter as a prospective client.
Can a law firm take the other side after a caller talked to its AI?
Yes, if the bot never collected information that could be significantly harmful to the caller. If it did, Rule 1.18(d) allows the representation with informed consent from both the affected client and the prospective client, confirmed in writing, or through the screening alternative. A solo has no colleague to screen, so collecting less up front is the stronger protection.
How long do we keep old AI intake scripts?
The conservative answer is 3 years after last regular use. Nevada Rule 7.2A(c)(4) requires a copy of information provided under that rule to be kept that long. If your scripts or follow-up messages fall within it, every version is a retained record. Save each with a date, since many platforms overwrite prompts in place.
What to do this week
- Pull ten recent intake transcripts and mark each field against the audit table. If any contain the caller's own account of what happened, your bot is collecting facts before a conflict check, and stage one needs to stop at names.
- Add the limiting script in English and Spanish, and log consent. Have the responsible lawyer approve the wording, name that lawyer in follow-up texts, and store each answer with a timestamp and script version.
- Lock down where transcripts land. Move them out of shared inboxes and firm-wide search into a store with per-matter access and a log, then ask your vendor in writing whether its AI features read across all firm conversations.
If you want a second set of eyes, our free 15-minute audit walks your intake flow against this checklist; start at AI for law firms in Las Vegas.
Drafted with AI assistance, researched, edited, and fact-checked by Elias Musleh on November 14, 2026.
Free 15-min automation audit
Want this handled for you?
We build the automation, run it, and hand you the numbers. Book a fifteen minute call and we will tell you straight whether it is worth doing for your business.
Or reach us directly: vendors@vegasbusinessai.com 702.826.9055