AI drafting tools are now part of many law firm marketing workflows. They outline blog posts, rewrite practice area pages, and summarize statutes. The output usually reads smoothly, and that can be a problem. Polished writing can still cite a wrong deadline, misstate a statute, or invent a case that doesn’t exist.
In 2026, the State Bar of California proposed amendments to the Rules of Professional Conduct that would address artificial intelligence directly. The proposal is aimed at how lawyers practice law, not at marketing. Still, its core message applies to law firm content: AI output needs meaningful human review before anyone relies on it.
Here is where the proposal stands, what it would require of California lawyers, and what firms can take from it when building their own AI-assisted content workflows.
Where Things Stand With the Amendments to California Bar Rules
The amendments are proposed, not in effect. In August 2025, the California Supreme Court sent a letter to the State Bar asking it to consider incorporating the principles of its 2023 Practical Guidance for the Use of Generative Artificial Intelligence in the Practice of Law into the California Rules of Professional Conduct.
The State Bar’s Standing Committee on Professional Responsibility and Conduct (COPRAC) developed the proposed amendments in response. COPRAC revised the proposal after the first public comment period closed on May 4, 2026. The modified version underwent a second round of public comment, which closed on August 6, 2026. Until the amendments are finalized and approved, they are not enforceable rules.
Separately, the State Bar approved its 2026 guidance on the use of generative AI. That guidance helps explain how lawyers’ existing professional responsibilities apply to AI, even while the proposed amendments remain under consideration.
What the Proposed Amendments Cover
The proposal would amend six existing rules governing lawyers’ professional responsibilities. The changes would clarify how those duties apply when lawyers use AI.
Competence (Rule 1.1). The proposal would add a definition of artificial intelligence to the existing comment on technology. It would also add a new comment stating that when a lawyer uses technology, including AI, the lawyer must exercise professional judgment over all aspects of that use, including the inputs and the outputs. When citing legal authority to a client or other parties, the lawyer would also have to verify that the authority is accurate and exists.
The “inputs and outputs” framing is the part most relevant to content workflows, because it treats AI as a tool whose use, from what goes in to what comes out, calls for professional judgment.
Client communication (Rule 1.4). The proposal would add a comment clarifying that the duty to keep clients reasonably informed includes evaluating what the lawyer needs to communicate about the use of technology, including AI. That duty runs through the representation and depends on the facts, including how novel the technology is, its risks and benefits, the scope of the work, and the client’s sophistication.
Confidentiality (Rule 1.6). The proposal would clarify that “revealing” confidential information includes exposing it to technological systems, including AI tools, where that creates a substantial risk the information may be used in a way that conflicts with the lawyer’s confidentiality duty.
In judging that risk, a lawyer should consider how the technology works, including whether data is retained or used for model training, what security measures are in place, and whether unauthorized users or third parties could access the information.
Candor toward the tribunal (Rule 3.3). A lawyer’s duty of candor would expressly include verifying that cited authorities exist and are not fabricated, misstated, or taken out of context, including authorities generated or assisted by AI. The proposed amendment addresses the risk of hallucinated citations that have shown up in court filings.
Supervision by managers (Rule 5.1). Managerial lawyers would need to make reasonable efforts to establish internal policies and procedures governing AI use.
Supervision of nonlawyer assistants (Rule 5.3). Lawyers would need to give nonlawyer assistants appropriate instruction and supervision on all ethical aspects of their work, including the use of AI.
What This Could Mean for Content and Marketing
Most of these rules govern legal services, not blog posts. A firm’s website content isn’t a court filing, and a practice area page isn’t legal advice to a client. But the proposal points in one consistent direction: fluent output is not verified output, and a responsible human has to stand behind it.
That has practical consequences for firms that publish AI-assisted content, because legal content is full of claims that can become outdated or simply be wrong:
- Statutes and code sections
- Case references and holdings
- Filing deadlines and statutes of limitations
- Procedural guidance, such as what happens after a claim is filed
- Jurisdiction-specific statements, including anything that varies between California and other states
An AI tool can produce all of these confidently and incorrectly. It may blend rules from several states, cite a superseded statute, or invent a case that sounds real. None of that shows up as a grammar problem, so a quick read for style won’t catch it.
The supervision rules add a second lens. Many firms don’t produce content alone. Agencies, freelancers, and internal marketing teams often draft it, and AI tools may be involved at any stage. The proposed Rule 5.1 and 5.3 comments emphasize internal policies around AI use and appropriate supervision of nonlawyer assistants. Firms should know who is responsible for drafting, verifying, reviewing, and approving content—and who is accountable if an error slips through.
AI-Assisted Content Is Already Part of the Law Firm Search Landscape
In our study of AI-generated law firm content, we analyzed 4,143 firm-owned pages and 2,506 high-ranking organic results. We found that 22.15% of the firm-owned pages showed AI-generation signals, and 18.99% of top-eight results did.
We didn’t find evidence that AI use alone determines whether content ranks. If firms are going to use these tools, the important operational question becomes: What happens between AI output and publication? The proposed California rules point to a similar need for human oversight when AI is involved.
Practical Steps Law Firms Can Take Now
Whether your practice is based in California or not, the principles behind the proposed rules offer a useful framework for managing AI-assisted content. Here are seven practical steps firms can take now to build more oversight into their content workflows.
1. Define Who Does What
Each piece of content should have four clearly defined responsibilities, even if one person handles more than one:
- Drafting: who creates the first version, whether a writer, an AI tool, or both
- Fact-checking: who verifies every factual and legal claim against a primary source
- Attorney review: who confirms that the legal substance is accurate and appropriate for the firm’s jurisdiction
- Final approval: who has authority to publish
Write this down and apply it to agencies and freelancers as well. If an outside partner drafts your content, the firm should know how that content is checked and who signs off before it goes live.
2. Treat Legal Claims as Things to Verify, Not Proofread
Verification means checking against the primary source, not asking the AI tool whether it is sure. Giving reviewers a checklist can help. For each page, someone should confirm:
- Every statute, rule, and code section is cited correctly and is current
- Every case reference actually exists, says what the content claims, and is still good law
- Deadlines and time limits match the current law in the relevant jurisdiction
- Procedural descriptions reflect how the process actually works
- Statements about state laws don’t “borrow” from another state’s rules
3. Set an Internal AI Policy for Your Firm
Proposed Rule 5.1 would require managerial lawyers to make reasonable efforts to establish internal policies and procedures governing AI use. A policy is useful whether the proposed amendment is ultimately adopted or not. A workable one covers:
- Which AI tools are approved for content work, and for which tasks
- What may never be entered into a prompt, including client names, case facts, privileged material, and internal strategy
- Which types of content require attorney review before publication
- How AI use is documented
The confidentiality issue deserves particular care. If marketing teams use client stories, intake notes, or case results as inputs for AI-assisted content, firms need to understand what the tool does with that information before it is entered.
4. Extend the Policy to Vendors
If an agency or freelancer works on your content, ask direct questions. Do they use AI in drafting? Which tools? What is their fact-checking process? Who reviews legal claims before delivery? A good partner will have clear answers. A vague answer is a warning sign.
AI oversight is only one part of the review process. Our guide on bar-compliant law firm website content covers the separate advertising rules that can apply to attorney claims, disclaimers, and testimonials. Those requirements should be part of the same review process.
5. Prioritize by Risk
Not every page carries the same risk if something is wrong. A page explaining a statute of limitations or a claims process deserves more scrutiny than a post about a community event. Start your review with high-stakes content: practice area pages, FAQ pages, and any content that gives procedural or deadline-related information. Then work outward.
6. Keep Clear and Dated Records
Note who drafted, checked, and approved each piece, and when. If a question comes up later about how a page was produced or why it said what it said, a simple record answers it.
7. Revisit Older Content Drafted or Published with AI Assistance
Periodically audit existing AI-assisted content, starting with pages that state legal rules or deadlines. That review can catch outdated or unsupported claims before a prospective client or a bar reviewer does.
AI-Assisted Content Still Needs Human Oversight
The proposed amendments are still just proposals, and they are aimed at how lawyers practice, not how firms market. But they reflect a consistent principle: lawyers should independently verify AI output and put oversight structures around it. Law firm content deserves the same discipline.
AI can still play a useful role in content strategy, drafting, and editing. The mistake is treating a draft that reads well as a draft that is ready to publish.
If your firm wants help building a content review workflow with clear roles for drafting, fact-checking, attorney review, and approval, MeanPug Digital can help. Contact us to talk it through.
