Compliance and Privacy10 min read
ABA Formal Opinion 512: AI Intake Duties
What ABA Formal Opinion 512 means for AI intake at Nevada law firms, with a duty table, consent limits, vendor checks, and records to keep.
ABA Formal Opinion 512 puts duties before deployment
ABA Formal Opinion 512 tells lawyers using generative AI to address competence, confidentiality, communication, supervision, and reasonable fees before treating a tool as ready for client work. For a Las Vegas law firm, that means evaluating the proposed intake workflow and its disclosures before entering protected information, with informed consent required for the self-learning tools described in the opinion. The ABA opinion issued July 29, 2024 is the newest dated authority supplied for this analysis, so its guidance predates 2025; Nevada's professional conduct rules govern Nevada lawyers.
The distinction matters. An ABA formal opinion is persuasive guidance, not a binding Nevada rule or a product certification. This article is general information, not legal advice, and translates the supplied opinion extracts into proposed operating controls rather than claiming every control is expressly required.
Start with the intended activity. Answering office-hours questions, collecting a caller's account of an accident, and summarizing an existing client's litigation history expose different information. A purchasing decision becomes an ethics decision when the firm chooses what information the tool receives, where it travels, and who relies on its output.
Translate each duty into an approval record
The table separates the opinion's words from our implementation recommendations. The competence and communication quotations are short excerpts from its summary holding, not complete explanations of those duties. The other quotations come from its confidentiality, supervision, and fees discussions in the opinion text.
| Duty | What Formal Opinion 512 says verbatim | What it means for an AI phone or intake tool | Firm action item | Evidence to keep on file |
|---|---|---|---|---|
| Competence 1.1 | "to provide competent legal representation" | The firm needs a basis for deciding whether generated intake information is dependable enough for its intended use. | Test summaries, escalation, and factual capture with fictional matters before approving the workflow. | Test cases, observed failures, corrections, and the approving lawyer's scope decision. |
| Confidentiality 1.6 | "a client's informed consent is required prior to inputting information relating to the representation into such a GAI tool." | The quoted requirement concerns the self-learning tools described in the opinion. Assess disclosure risks before routing client information into the system. | Map information flows and determine whether the proposed use requires informed consent. | Data-flow record, applicable vendor terms, risk explanation, and consent record where required. |
| Communication 1.4 | "to communicate with clients" | Decide what the client must understand about the proposed use and its consequences. | Have counsel approve explanations tied to the actual workflow, including any consent discussion. | Approved explanations, delivery records, client questions, and responses. |
| Supervision 5.1 and 5.3 | "Managerial lawyers must establish clear policies regarding the law firm's permissible use of GAI, and supervisory lawyers must make reasonable efforts to ensure that the firm's lawyers and nonlawyers comply with their professional obligations when using GAI tools." | Staff need defined permissions and an accountable supervisor. | Name a responsible lawyer, set permitted uses, and train everyone operating or reviewing the tool. | Policy, training record, review assignments, and incident decisions. |
| Fees 1.5 | "lawyers who bill clients an hourly rate for time spent on a matter must bill for their actual time." | Automation does not create billable lawyer hours that nobody worked. | Review timekeeping and any proposed client charge for the tool. | Actual time entries, fee terms, expense explanation, and billing review. |
| Vendor diligence 5.3(b) | "These concepts also apply to GAI providers and tools." | The opinion extends its outsourcing diligence discussion to generative AI providers. | Investigate security, credentials, personnel practices, confidentiality terms, conflicts processes, and legal remedies. | Dated vendor responses, supporting documents, contract terms, and unresolved concerns. |
Keep this record specific to a workflow and configuration. Approval for a public-information assistant should not silently become approval for recording client narratives or generating legal recommendations. A new integration can change the exposure even when the vendor name stays the same.
These are useful acceptance criteria when evaluating AI for Las Vegas law firms. A demonstration that sounds natural answers a usability question; the approval record answers who considered the professional obligations and on what evidence.
Boilerplate cannot supply the missing consent discussion
The confidentiality passage concerns self-learning tools whose outputs could directly or indirectly disclose information relating to a representation. In that setting, the opinion requires informed consent before inputting that information. Read the condition alongside the requirement in the confidentiality discussion on page 7, rather than turning it into a universal rule about every software product.
Consent language that fails "To obtain informed consent when using a GAI tool, merely adding general, boiler-plate provisions to engagement letters purporting to authorize the lawyer to use GAI is not sufficient." Source: Formal Opinion 512, confidentiality discussion.
Our example of the rejected approach is an engagement clause saying the firm may use technology, including AI, to deliver services. That sentence leaves the client without an explanation of what information enters the tool or what exposure the firm identified. Signing it does not supply the missing explanation.
Build the conversation from the actual proposed use. Describe what information would be processed, what the tool would do, and the material risks the lawyer has identified. Offer a practical route for handling the matter if the client declines that use, and preserve the explanation and response together.
Timing is especially important for phone intake. If sensitive information reaches the model before the firm determines whether consent is necessary, a later signature cannot make that input occur after consent. Design the opening workflow so the firm can resolve the issue before the relevant disclosure.
A general announcement that a caller is speaking with AI is not automatically informed consent for the disclosure risks discussed here. Likewise, this ethics analysis does not settle recording consent. Treat those as separate questions requiring their own review.
Supervision must reach the people operating the workflow
The opinion expressly includes training in supervisory obligations. Its supervision discussion on pages 10 and 11 calls for training subordinate lawyers and nonlawyers in the ethical and practical use of relevant tools and their risks. A vendor tutorial about buttons and settings does not, by itself, establish what the firm permits.
For a small practice, assume a lawyer supervises intake staff and reviews generated summaries. The practical policy can be short: identify approved tasks, prohibited inputs, required review, and escalation routes. Staff should know what to do when a caller volunteers information outside the approved intake scope.
For a larger practice, assume reception, practice groups, and billing use different systems. The handoffs become part of supervision because an approved intake summary may be copied into another tool with different terms. Assign responsibility for each handoff instead of assuming the lawyer reviewing the final file controls every earlier disclosure.
Nevada supplies the local hook. Nevada Rule of Professional Conduct 5.3 requires reasonable efforts by managers and direct supervisors to keep nonlawyer assistance compatible with lawyers' professional obligations. It also addresses responsibility for ordering or knowingly ratifying conduct and for failing to take reasonable remedial action in specified circumstances.
Vendor diligence should therefore produce answers, not just a badge or sales assurance. The opinion names credentials, security policies, hiring practices, confidentiality agreements, conflicts screening, and access to a legal forum for contractual violations. Ask how each relevant consideration applies to the provider and document gaps that affect approval.
For an AI receptionist in Las Vegas, test a caller who changes the subject from scheduling to legal advice. Our recommended acceptance test is whether the system stays within its approved scope and reaches the designated human. Keep the result, including failures, rather than recording only that a demonstration occurred.
Faster drafting changes the billing review
The opinion's billing example is concrete. In its fees discussion on page 12, a lawyer spends 15 minutes entering information to generate a pleading. The lawyer may bill that actual time and the actual time spent reviewing the draft for accuracy and completeness.
For our worked example, assume review takes another 15 minutes. The total is 15 minutes of input plus 15 minutes of review, equaling 30 minutes of actual lawyer time. The review duration is our assumption; the opinion supplies the input-time example and the actual-time principle.
Now assume a larger practice uses the same process on separate matters. The rule does not change with practice size: record the work actually performed on each matter. Time saved compared with a previous manual process is not additional time expended on a client's behalf.
Flat fees need attention too. The opinion says it may be unreasonable to charge the same flat fee when generative AI makes tasks much faster, and states that a fee for little or no work is unreasonable. That is a reason to review the arrangement and work performed, not a formula requiring every efficiency gain to produce an identical discount.
Separate professional fees from tool expenses. Before charging a client for a generative AI tool as an expense, the opinion says the lawyer must explain the basis, preferably in writing. Decide whether and how the firm will charge before staff begin allocating software costs to matters.
This analysis does not approve every intake system
Formal Opinion 512 addresses generative AI. A fixed scheduling form, a scripted phone menu, and a generative assistant may have different functionality and risks. Do not infer that the opinion's particular self-learning-tool consent passage applies identically to all of them, or that ordinary confidentiality obligations disappear outside that passage.
The strongest objection is that a vendor promises not to train on firm data. That promise matters, but it answers only part of the workflow review. Investigate retention, access, onward processing, and integrations before deciding what the promise establishes for the intended use.
This playbook also does not resolve every issue involving prospective clients, conflicts, privilege, recording, or court-specific requirements. An intake system may encounter those issues before an engagement exists. The supplied extracts support the duties discussed here, not a complete legal clearance for deployment.
Nor does this article establish that the cited authorities remained unchanged through publication. The opinion is dated; the supplied Nevada rule extract has no separate verification date. Counsel should check the applicable authority and current vendor terms before relying on an implementation decision.
For custom AI assistants, customization gives the firm design choices rather than an ethics exemption. A narrower information boundary may change the risk assessment. Document why the chosen boundary is adequate and what change would require reconsideration.
Questions owners ask
Is ABA Formal Opinion 512 binding on Nevada lawyers?
No. ABA formal opinions are persuasive guidance, while Nevada lawyers are bound by Nevada's professional conduct rules. The opinion helps identify how established duties apply to generative AI. The Nevada nonlawyer-assistance rule supplies a local basis for assessing supervision, but the firm still needs a workflow-specific evaluation under its applicable rules.
Can our engagement letter authorize all AI use?
General boilerplate is not sufficient informed consent under the opinion's quoted guidance. Where informed consent is required, the explanation must address the proposed use rather than merely announce that the firm uses technology. Keep the explanation and client's response together, and resolve the consent issue before the information enters the tool covered by that requirement.
Can lawyers bill for time saved by AI?
Lawyers billing hourly must bill actual time, including time actually spent preparing inputs and reviewing output. They cannot bill hypothetical hours that a slower process would have required. Flat fees also remain subject to reasonableness, and charging a client for the tool as an expense requires explaining the basis, preferably in writing.
What to do this week
- Map the proposed intake path. Write down what callers provide, which systems receive it, where it is stored, and who can access it. Use fictional information while investigating. Have the responsible lawyer decide the permitted scope and whether consent is required before client information enters the relevant tool.
- Build the approval file. Complete the duty table for your actual configuration, obtain vendor responses, and preserve the terms reviewed. Assign a supervisor and train staff on permitted inputs, human escalation, and output review. Record unresolved issues as deployment blockers when they prevent a supported decision.
- Run a controlled rehearsal. Test routine scheduling, unexpected sensitive disclosures, inaccurate summaries, and requests for legal advice. Correct failures, document the review, and check billing treatment before launch. Identify which vendor or workflow changes will trigger another review so approval remains tied to the system the firm actually uses.
For a second look at your proposed workflow, request a free 15-minute audit.
Drafted with AI assistance, researched, edited, and fact-checked by Elias Musleh on November 2, 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