State bar guidance on AI in 2026 is clearer than most attorneys assume. The American Bar Association issued Formal Opinion 512 in July 2024. It established that lawyers must understand AI's capabilities and limitations and verify all AI-generated output. Since then, California, Florida, New York, New Jersey, Illinois, Texas and Oregon have all published AI guidance for lawyers. The operating principles have converged. "The rules are unclear" no longer describes where things stand.
This article walks through what state bar guidance actually says about AI case intake, what the ABA's 2024 formal opinion requires, the seven-point compliance checklist that maps to the live guidance, and a state-by-state look at where the rules are currently most developed.
Which rules govern AI case intake for lawyers?
Three layers govern AI intake. ABA Formal Opinion 512 (July 2024) sets the national baseline. State guidance from California, Florida, New York, New Jersey, Illinois, Texas and Oregon applies it locally. Underneath both sit the existing Model Rules on competence, confidentiality, supervision and advertising, which treat AI as a tool the lawyer remains responsible for.
Three sources now anchor most of the ethics conversation around AI intake in US law firms:
- The American Bar Association's Formal Opinion 512 on generative AI tools (July 2024), which sets the national baseline.
- State bar opinions and ethics committee guidance. California's COPRAC Practical Guidance for the Use of Generative Artificial Intelligence was the first state framework to apply existing conduct rules to AI. Florida's Ethics Opinion 24-1 is the one that addresses intake chatbots directly. New York has its NYSBA Task Force on AI report. Texas Opinion 705 (February 2025) is specific about self-learning tools that retain what you type into them. Oregon's Formal Opinion 2026-208 (February 2026) is the newest, and it covers AI agents in client intake. Illinois and New Jersey add court-issued guidance.
- ABA Model Rules 1.1 (competence), 1.6 (confidentiality), 5.1 / 5.3 (supervision), 7.1 (advertising), and 8.4 (misconduct) — applied to AI as a "tool" rather than as a separate category of regulated activity.
The federation-level message across all of these sources is consistent: AI is a tool. The lawyer is responsible. The client must know they are interacting with AI when they are. Everything else is implementation detail.
Can a law firm use AI for client intake?
Yes, with conditions. None of the guidance reviewed here bans AI intake. It permits AI to capture contact details, screen practice area and conflicts, and book consultations. The conditions: the prospect is told they are talking to an AI, and a licensed attorney gives all legal advice and decides whether to take the matter.
The most common misunderstanding in 2026 is that AI in client intake is broadly prohibited. It is not. Florida's Opinion 24-1 is the most explicit. A lawyer using a generative AI chatbot for intake must inform prospective clients that they are communicating with an AI program. Oregon's Formal Opinion 2026-208 answers the intake-agent question "yes, qualified." The lawyer must monitor the agent so it does not appear to form an attorney-client relationship or guarantee results. Within those limits, AI may handle:
- Initial intake — capturing names, contact info, jurisdiction, basic facts about the matter.
- Practice-area screening — confirming whether the matter falls within the firm's accepted practice areas.
- Conflict-of-interest screening — running the prospective client and adverse parties through the firm's database.
- Scheduling — booking a consultation with a licensed attorney.
- Confirmation messaging — sending automated text or email reminders.
Some things AI may not do in any state. It may not render legal advice or quote case outcomes. It may not set fees in a way that amounts to a definitive engagement. And marketing may not claim the AI is superior to other tools without objectively verifiable evidence. The boundary is between intake and counsel: intake is fine with disclosure, counsel is not.
Convergent guidance across the ABA and state guidance (CA, FL, NY, NJ, IL, TX, OR) as of 2026
Always verify the rules in your specific jurisdiction. State bar guidance evolves — this summary represents convergent guidance as of early 2026 and is not legal advice.
Why do some bars treat AI intake as high-risk?
The risk is unsupervised intake, not intake itself. Guidance flags AI that can give legal opinions, promise results or appear to accept a representation. When the AI only collects facts, screens and books, and an attorney reviews every matter before engaging, the setup matches what published guidance permits.
Some state bar guidance has labeled client intake as a heightened-risk area. This is sometimes summarized as "intake is a Red Light." The framing is misleading without context. What bar associations have actually flagged is unsupervised, ungated AI intake. That is intake where the AI can give legal opinions or commit the firm to a representation. That configuration is the prohibited one.
The compliant configuration is structurally different. AI handles the intake conversation, captures the data, screens for conflicts, and books the consultation. A licensed attorney remains the only party giving legal advice and the only party able to accept a representation. The lawyer reviews the intake brief before the consultation, conducts the consultation themselves, and decides whether to engage. Under that configuration, the guidance reviewed in this article permits AI intake.
What does a compliant AI intake setup need?
Seven controls: disclose the AI on first contact, block legal advice, screen conflicts before intake ends, tell prospects how their data is handled, verify AI output before it leaves the firm, document attorney supervision under Rules 5.1 and 5.3, and keep AI marketing claims objectively verifiable under Rule 7.1.
For solo and small firms deploying AI case intake in 2026, the practical compliance checklist generally looks like this:
- Disclosure on first contact. The AI identifies itself as an AI assistant within the first turn of the conversation, in plain language. ("Hi, I'm an AI assistant for [Firm]. I'll take some initial information so an attorney can follow up with you. Anything you share is treated as confidential.")
- No legal advice from the AI. The intake script is hard-gated against opinions on the merits, recoverability, settlement value, or jurisdiction-specific legal questions. Those queries trigger an immediate offer to schedule with the attorney.
- Conflict-screening before intake completes. Names, opposing parties, and matter type are checked against the firm's conflict database. If a conflict surfaces, the conversation ends and the prospective client is notified.
- Confidentiality and data-handling disclosure. The prospective client is told what data is being captured and how it is stored. Any AI vendor used for intake should sign a business associate agreement (or equivalent confidentiality contract) and the data should not be used to train external models.
- Verification of AI output before any external use. No AI-generated text — citations, summaries, draft letters — leaves the firm without a licensed attorney verifying it. This applies to anything customer-facing, not just court filings.
- Supervision under Model Rules 5.1 and 5.3. The attorney supervising the AI tool is responsible for its outputs as if it were a non-lawyer assistant. Documenting that supervision (regular review, sample audits) is what the published guidance expects.
- Advertising compliance under Model Rule 7.1. Marketing copy referencing AI cannot make superiority claims that are not objectively verifiable. "Our AI is the best" is out. "Our AI handles after-hours intake 24/7" is fine because it is verifiable.
Which states have issued AI ethics guidance for lawyers?
At least seven states have published guidance relevant to AI intake: California (2023), Florida and New Jersey (January 2024), New York (April 2024), Illinois (effective January 2025), Texas (February 2025) and Oregon (February 2026). All of them apply existing conduct rules rather than writing new ones. The table summarizes each.
If your firm operates in multiple jurisdictions, the rules are not yet uniform across all 50 states. Here is the snapshot, with the practical implication for AI intake:
| Jurisdiction | Guidance and date | What it means for AI intake | Source |
|---|---|---|---|
| ABA (national) | Formal Opinion 512, July 2024 | Applies competence, confidentiality, communication, supervision and fee rules to generative AI | americanbar.org |
| California | COPRAC Practical Guidance, 2023 | Existing conduct rules apply to AI, with weight on confidentiality and supervision | calbar.ca.gov |
| Florida | Ethics Opinion 24-1, January 2024 | Intake chatbots must disclose they are AI and must not give legal advice | floridabar.org |
| New Jersey | Supreme Court preliminary guidelines, January 2024 | Existing Rules of Professional Conduct apply unchanged to AI use | njcourts.gov |
| New York | NYSBA Task Force report, April 2024 | Recommends telling clients when AI tools are used in their matters | nysba.org |
| Illinois | Supreme Court AI policy, effective January 2025 | Conduct rules apply fully; lawyers must review AI output; no disclosure required in pleadings | Illinois Commission on Professionalism |
| Texas | Opinion 705, February 2025 | Do not enter confidential data into tools that may expose it without client consent; verify output | legalethicstexas.com |
| Oregon | Formal Opinion 2026-208, February 2026 | AI intake agents allowed, qualified: supervise the provider and prevent implied representation or guarantees | osbar.org |
"The compliant pattern is well-defined: disclose, escalate legal advice to humans, screen for conflicts, supervise the tool, and verify the output. Inside that pattern, AI case intake is one of the most practical operational upgrades available to solo and small firms."— Editorial summary
California
The State Bar of California issued comprehensive Practical Guidance for the Use of Generative Artificial Intelligence in the Practice of Law. The guidance closely tracks the ABA framework, with particular emphasis on confidentiality (Rule 1.6) and supervision (Rule 5.3). California is also notable for its consumer-protection-flavored disclosure expectations — when a client interacts with a chatbot, the AI's identity must be clear and the client must know how their data is handled. AI intake configured per the 7-point checklist above is permitted.
New York
The NYSBA Task Force on Artificial Intelligence report was approved by the House of Delegates on April 6, 2024. It recommends telling clients when AI tools are used. New York's approach emphasizes confidentiality and the duty of supervision under Rule 5.3. In effect, the AI is treated like a non-lawyer staff member whose work the supervising attorney must oversee. Nothing in it bars AI intake that follows the checklist above.
Florida
Florida Bar Ethics Opinion 24-1 explicitly addresses generative AI in legal practice. It recommends obtaining the affected client's informed consent before using a third-party GenAI tool that would involve disclosing confidential information. Lawyers must also verify the accuracy of AI-generated output. For intake specifically, the disclosure-and-supervision framework applies.
Illinois
The Illinois Supreme Court's policy on artificial intelligence took effect on January 1, 2025. It says the Rules of Professional Conduct apply fully to AI use and holds lawyers accountable for reviewing AI-generated work. It does not require disclosure of AI use in pleadings. Illinois firms should follow the same 7-point checklist with attention to the state's specific advertising rules under Illinois Rule 7.1.
New Jersey
The New Jersey Supreme Court issued preliminary guidelines in January 2024 affirming that lawyers may use AI tools, with the standard caveats: maintain confidentiality, verify accuracy, and supervise the work product. New Jersey's framework is fully compatible with the AI intake configuration described in this article.
For states without specific guidance, the ABA framework plus existing Model Rules apply. As more states issue guidance, expect convergence with the framework outlined in California, Florida, and New York.
What should a compliant AI intake tool include?
Look for five defaults: an AI disclosure built into the first message, automatic escalation of legal-advice questions to a human, conflict-screening fields mapped to your matter database, full transcript logging, and a written vendor commitment that intake data stays confidential and is not used to train outside models.
Most US firms deploying AI intake in 2026 are not building it themselves. The tools that map cleanly onto the compliance checklist above tend to share a few features:
- The disclosure prompt is built into the AI's first turn — not an optional configuration that can be skipped.
- Hard-coded escalation triggers move legal-advice questions to a human queue automatically.
- Conflict-screening hooks are in the intake form itself, with field mapping to the firm's matter database.
- All conversation transcripts are logged for the lawyer's review and for state bar audit if requested.
- The AI vendor signs a vendor confidentiality / BAA equivalent and confirms in writing that intake data is not used to train external models.
Firms running TheBigBot's legal CRM ship with these defaults configured at delivery — disclosure on first turn, hard escalation on legal-advice queries, conflict-screening prompts, full transcript logging, and a vendor confidentiality posture that is reviewable by the firm's compliance partner. The AI receptionist, intake CRM, scheduler, and review-harvesting workflows live in one login that is typically live in 3 days.
Which law firms benefit most from AI intake?
Firms whose leads arrive when nobody can answer. That means personal injury, family law and immigration practices with heavy evening and weekend calls. It also means firms spending heavily on ads with one person on intake, and firms serving clients who prefer a language other than English. The gap is capacity, not compliance.
The compliance question is settled enough that the practical question — "where does this actually move the needle?" — is now the more useful one. Across US solo and small firms in 2026, the highest-impact AI intake deployments share three properties:
- Practice areas with high after-hours call volume. Personal injury (post-accident calls), family law (private/evening calls), immigration (cross-time-zone calls). These are the areas where the 24/7 capture is decisive — and where the existing answering-service or voicemail option leaks the most leads.
- Firms with high marketing spend and uneven intake capacity. A firm running $10K+/month in Google Ads and routing calls through a single paralegal is unlikely to capture all the leads it is paying for. AI intake reduces the staffing bottleneck without adding headcount.
- Multi-language client populations. Spanish-speaking, Mandarin-speaking, Arabic-speaking, and other non-English-primary client populations are systematically underserved by English-only answering services. AI intake handles the language switch natively.
What mistakes put AI intake at risk with the bar?
Three mistakes cause most of the risk: making the AI disclosure optional, letting the AI quote a fee for a specific matter, and not keeping transcripts. Each one removes a control the guidance relies on. The last leaves you unable to show what the AI told a prospective client if a complaint arises.
Three common implementation errors that turn an otherwise-compliant AI intake into a state bar risk:
- Skipping the disclosure prompt because "it slows down the conversation." Florida's opinion requires it outright, and it is the safe default everywhere else. Configure it to land in the AI's opening line and do not let it be optional.
- Letting the AI quote fees for specific matters. Generic "our consultations are $X" or "our typical retainer range is $Y to $Z for this type of matter" is fine. "Your case will cost $15,000" is the line that crosses into engagement territory and should always be a licensed attorney's call.
- Not logging conversations. If a state bar audit ever asks "what did the AI say to this prospective client" and you cannot produce a transcript, the supervision-record question becomes harder to answer. Full transcript logging is a baseline requirement, not a nice-to-have.
Frequently Asked Questions
Does my state bar require attorneys to take CLE on AI ethics?+
Check your own state's rules. Comment 8 to Model Rule 1.1 already expects lawyers to keep up with the benefits and risks of relevant technology, and some states, such as Florida, require technology CLE credits. CLE programs on AI ethics are widely available. Confirm your state bar's current CLE requirements before assuming any specific obligation.
Can an AI handle a fee quote during intake?+
Generally no. Quoting a specific fee for a specific matter often crosses into the "definitive engagement terms" zone where state bar rules expect a licensed attorney to be the speaker. Most compliant AI intake systems handle the practice-area-fit conversation, but defer the specific fee discussion to the attorney consultation. Always verify the rules in your jurisdiction.
What if my state hasn't issued AI ethics guidance yet?+
The national baseline is ABA Formal Opinion 512 (July 2024) plus the existing Model Rules. Most state bars treating AI as a "tool" — rather than a separate category — apply existing rules on competence, confidentiality, supervision, and advertising. When state-specific guidance does come out, it generally aligns with the ABA framework. Following the ABA framework is a reasonable default until your state issues specific guidance, and reviewing your bar's website periodically is the standard practice.
Do I have to disclose AI use to existing clients, or just prospective ones?+
Disclosure obligations apply to both, but the form differs. Prospective clients in an intake conversation should be told at the start that they are talking with an AI. Existing clients should generally be informed if AI is being used to handle aspects of their matter (drafting, research, scheduling) per Model Rule 1.4 (communication). Specific disclosure expectations vary by jurisdiction and by the type of AI use.
Where is the line between intake and legal advice?+
"Tell me what kind of case I have" is the line. AI may confirm a matter falls within a practice area the firm accepts (intake). AI may not characterize the merits, jurisdiction-specific legal classification, or potential outcomes (legal advice). When the conversation drifts toward "what should I do" or "how strong is my case," compliant AI intake systems escalate to a human attorney rather than respond.
Does my malpractice carrier care about AI intake?+
Ask your carrier directly, because underwriting questions vary by carrier. If AI use comes up, the useful answer is not "we do not use AI" but "we use AI with documented supervision." Having the 7-point checklist written into your firm's AI policy gives you that answer in writing.
Can the AI receive privileged communications?+
Yes, if structured as a confidential agent under appropriate vendor agreements. The AI vendor functions analogously to a non-lawyer assistant under Model Rule 5.3 — bound by confidentiality, supervised by an attorney, and contractually prohibited from disclosing or training on the privileged content. The BAA / vendor confidentiality agreement is the contractual mechanism that makes this work.
Does using AI intake affect attorney-client privilege?+
Properly configured, no. The standard analysis treats the AI as a confidential agent of the firm, similar to a paralegal or receptionist. Privilege attaches to communications made for the purpose of obtaining legal services, regardless of whether the initial intake is taken by a human or an AI assistant — provided the standard confidentiality protections (vendor BAA, no external training, secure storage) are in place. Always confirm the analysis with your jurisdiction's specific privilege rules.
The bottom line
The state bar rules on AI in client intake have stabilized enough in 2026 that compliance is now a configuration question. The firms still treating "AI ethics" as a reason to keep using voicemail are not protecting themselves from a state bar action — they are protecting their competition's market share. The compliant pattern is well-defined: disclose, escalate legal advice to humans, screen for conflicts, supervise the tool, and verify the output. Inside that pattern, AI case intake is one of the most practical operational upgrades available to solo and small firms. To see how intake tools differ on it, read our law firm intake software comparison.
If you'd like to see what an ethics-compliant AI case intake configuration — with disclosure prompts, conflict screening, attorney-only legal advice, full transcript logging, and review-harvesting wired into one login — looks like running on your firm's lead flow, book a 20-minute demo. We will walk through the configuration for your jurisdiction and your practice areas.
This article is for general informational purposes only and does not constitute legal advice. Reading this article does not create an attorney-client relationship. Ethics rules vary by jurisdiction and change frequently — consult your state bar's most current guidance and a licensed attorney in your jurisdiction before deploying AI tools in your practice.
References & sources
- American Bar Association, Formal Opinion 512: Generative Artificial Intelligence Tools (July 2024) — americanbar.org
- State Bar of California, “Practical Guidance for the Use of Generative Artificial Intelligence in the Practice of Law” (COPRAC) — calbar.ca.gov
- Professional Ethics Committee for the State Bar of Texas, Opinion 705 (February 2025) — legalethicstexas.com
- The Florida Bar, Ethics Opinion 24-1 (January 2024) — floridabar.org
- New York State Bar Association, Report and Recommendations of the Task Force on Artificial Intelligence (April 2024) — nysba.org
- New Jersey Supreme Court, Preliminary Guidelines on the Use of Artificial Intelligence by New Jersey Lawyers (January 2024) — njcourts.gov
- Illinois Supreme Court Commission on Professionalism, Illinois Supreme Court releases policy on AI — 2civility.org
- Oregon State Bar, Formal Opinion 2026-208 (February 2026) — osbar.org