Generative AI for lawyers: California November 2023, Florida January 2024, and Rules 1.1, 1.6, 5.5, and 7.1 in practice.
The California State Bar issued Practical Guidance for the Use of Generative Artificial Intelligence in November 2023. The Florida Bar followed with Advisory Opinion 24-1 in January 2024. Both apply existing ABA Model Rules to AI tooling: Rule 1.1 competence in the tool, Rule 1.6 confidentiality on inputs, Rule 5.5 unauthorized practice of law on outputs, Rule 7.1 accuracy on advertising claims about AI capability. The rules are not new. The application to generative models is new and the disciplinary exposure is real.
The state bar guidances
The California State Bar issued Practical Guidance for the Use of Generative Artificial Intelligence in the Practice of Law in November 2023. The guidance is the first U.S. state bar full framework on attorney use of generative AI. The Florida Bar followed with Advisory Opinion 24-1 in January 2024, addressing similar surfaces with Florida-specific operational notes. New York, Texas, and other state bars have since issued guidance of varying depth.
Both California and Florida align substantively. Attorney competence in the tool is required under Model Rule 1.1. Attorney supervision of AI output is required. Documented attorney review of client-facing AI-drafted content is required. Confidentiality protection under Rule 1.6 applies when feeding client information into a model. Accuracy under Rule 7.1 applies to any AI-generated advertising claims. The guidances differ in operational detail (which prompt structures count as confidential disclosure, what supervision documentation must capture) but converge on the core analysis: existing rules apply, and the attorney remains responsible for everything the AI produces under the attorney's name.
Rule 1.1 competence and AI tooling
Rule 1.1 requires a lawyer to provide competent representation to a client. Comment 8 to Rule 1.1 (the technology-competence amendment adopted by the ABA in 2012 and now adopted by most states) extends competence to the benefits and risks associated with relevant technology. State bar AI guidances apply the framework directly to generative models.
An attorney using a tool that hallucinates case citations without checking the output violates Rule 1.1 by failing to maintain competence in the tool's failure modes. Competence here is operational, not theoretical. The attorney must understand what the tool does, where it fails, and what verification each output requires. The 2023 federal sanctions case Mata v. Avianca (S.D.N.Y.) made the Rule 1.1 surface visible when attorneys submitted ChatGPT-fabricated case citations to a federal court and were sanctioned. The guidance landscape has firmed up around the principle that attorneys are responsible for verifying every AI-generated factual claim before it ships under the attorney's name.
Rule 1.6 confidentiality and prompt inputs
Rule 1.6 applies regardless of the tool. Feeding identifying client facts into a public model that retains training data on prompts breaches confidentiality unless the client has consented after consultation. The risk surface is the platform's training-data policy and the platform's retention policy on prompts and outputs.
The compliant architecture either uses an enterprise model with no-training contractual carve-outs (Microsoft Copilot for Microsoft 365, OpenAI Enterprise, Anthropic Claude for Work, Google Workspace's Gemini for Workspace) plus a documented confidentiality posture, or operates on anonymized inputs with identifying details stripped before the prompt. Both state bar guidances treat the choice of tool as a Rule 1.6 question, not just a Rule 1.1 question. The attorney must select a tool whose confidentiality posture clears the rule, then maintain operational discipline on what gets fed into the prompts.
Rule 5.5 and chatbots
Rule 5.5 prohibits the unauthorized practice of law. An AI-powered chatbot on the firm's website that answers user-submitted legal questions with specific legal advice routes directly into UPL territory because the chatbot is not licensed. The compliant design limits the chatbot to scheduling, intake-form completion, FAQ surfacing of published information, and routing to a licensed attorney.
The user-facing disclaimer must make clear that the chatbot is not providing legal advice and that no attorney-client relationship is formed by the interaction. The disclaimer text alone does not save a chatbot that actually answers legal questions with specific advice; the chatbot's behavior controls the analysis. The Florida Bar Advisory Opinion 24-1 addresses chatbot deployment directly with operational guidance on the disclaimer architecture and the bounded scope of permissible chatbot interaction.
Rule 7.1 and advertising claims about AI
Rule 7.1 applies to advertising claims about AI capability the same way it applies to every other advertising claim. The unjustified-expectation standard tests whether the claim could lead a reasonable person to form an expectation about a result the firm cannot guarantee.
Claims like "AI-powered legal research" or "AI-assisted contract review" are permissible when the firm actually uses the tools claimed, the attorney review chain is real and documented, and the claim does not imply outcomes the AI assistance does not actually produce. Claims that suggest the firm produces faster or cheaper results because of AI must be verifiable in the engagement record. The California State Bar guidance treats advertising claims about AI capability as a Rule 7.1 surface.
The attorney-review workflow
Both state bar guidances require a documented attorney-review workflow for AI-drafted client-facing content. The workflow is operational, not theoretical. The AI-drafted surface is logged. The attorney with subject-matter and jurisdictional competence reviews against the engagement letter scope. Edits are captured. Sign-off is recorded. The output reads as written by an attorney because an attorney reviewed it before it shipped.
For SEO copy on the firm's site the workflow applies to AI-drafted practice-area pages, FAQ answers, blog posts, and any other client-facing content. We build the workflow as part of the engagement. The workflow surfaces who reviewed which page, what changes were made, and what the final attorney sign-off captured. The audit trail is both a Rule 1.1 compliance signal and a defense if a complaint or malpractice claim ever surfaces.
Schema architecture for AI-generated content
AI-generated legal content benefits from the same technical structuring that any high-E-E-A-T content needs to rank. The LegalService schema wraps the firm. The Person schema nests inside it per attorney with full credentials. AI-drafted articles attribute through Article.author to the human attorney who reviewed and signed off, not to the AI system. The schema architecture lets Google's parser attribute the content's E-E-A-T to the reviewing attorney, not to a generative model with no professional credentials.
The structural pattern matters because Google's Reviews System and Helpful Content System both weigh author entity. AI-generated content attributed to a Person node with bar admissions, publications, and association memberships carries the credentialed attorney's E-E-A-T as the parser reads it. AI-generated content with no human author attribution or with the AI system as author carries no E-E-A-T signal and ranks accordingly.
If the firm wants the AI-content workflow built and the attorney-review chain documented for SEO content production, our AI content compliance service handles the architecture. The underlying schema work sits inside our law firm schema markup service.
For the broader Attorney SEO program the AI-content workflow sits inside, the homepage is the entry point.
Questions on AI compliance before the workflow build.
-
Why do you cite Model Rule and Florida subchapter numbers on every page?
Because the rule number is the rule. ABA Model Rule 7.1 governs case-result claims and testimonial structure. Florida Rule 4-7.13 requires objectively verifiable case results plus the proximity-placed disclaimer. Florida Rule 4-7.14 governs contingency fee disclosures. Generic compliance copy ('we follow bar advertising rules') without the rule number is what got the buyer burned by the prior agency. The citation tells the firm's bar counsel which rule applies where.
-
We are not licensed in Florida. Why does Florida law dictate our site?
National attorney SEO has to clear the strictest jurisdiction. Florida Subchapter 4-7 is the strict-state baseline: case-result disclaimer mechanics under Rule 4-7.13, fee-disclosure copy under Rule 4-7.14, PI direct-mail 30-day blackout under Rule 4-7.18. California aligned to the ABA framework in November 2018 so it clears with the same compliance pass. A handful of strict states (Florida primarily) set the lowest common denominator for content that needs to work everywhere.
-
Do we need bar pre-approval to iterate site copy?
Florida Rule 4-7.20 exempts law-firm website content from the 20-day pre-filing requirement that applies to TV, radio, and direct mail. Website copy still has to comply with the substantive advertising rules under Rules 7.1 through 7.5, but SEO content iterates without the regulatory bottleneck. The exemption is the structural reason a monthly retainer cadence is workable for a regulated practice.
-
Can you use AI to draft FAQ and practice-area pages?
Yes, through a documented attorney-review chain. The California State Bar issued AI guidelines in November 2023 and the Florida Bar followed in January 2024. Both require a documented attorney-review workflow for AI-drafted client-facing content. We build the workflow as part of the engagement: AI-drafted surface is logged, attorney reviews against the engagement letter scope, edits captured, sign-off recorded. The output reads as written by an attorney because an attorney reviewed it before it shipped.
-
We have five attorneys at one office. Can each attorney have a Google Business Profile?
No. Google's policy for legal services is one Google Business Profile per practicing office, not one per attorney. Florida Rule 4-7.12 reinforces the same constraint from the bar side: a bona-fide office is a physical location where the lawyer reasonably expects to furnish legal services on a regular basis. The compliant architecture is one GBP for the firm at the office address, with category specificity matching the firm's strongest practice area (Personal injury attorney, Criminal justice attorney, Family law attorney).
Rules 1.1, 1.6, 5.5, and 7.1 apply to AI tooling regardless of whether the firm has thought through the application. Book a diagnostic.
We read the firm's current AI tooling, the prompt confidentiality posture, the chatbot scope, the AI-content review chain, and the advertising claims about AI capability. The diagnostic comes back with the per-surface analysis and the workflow scope.