The Frontier, Revisited: California’s Generative AI Rules and Trusts & Estates Counsel, One Year Later

By Jeremy J. Ofseyer, Esq., and Sean Tannenbaum, Esq.**

A year ago, this newsletter examined the State Bar of California’s November 2023 “Practical Guidance for the Use of Generative Artificial Intelligence in the Practice of Law,” which applied existing professional duties to AI.ⁱ Since then, the California Supreme Court directed the State Bar to codify the 2023 Practical Guidance into binding rules; the State Bar’s Board of Trustees issued an updated 2026 Practical Guidance reaching agentic systems; the State Bar of California Standing Committee on Professional Responsibility and Conduct (COPRAC) proposed amendments to six rules and an advisory on AI hallucinations; and the State Legislature advanced a verification statute.

Although no known California authority has specifically addressed the use of AI by trusts and estates counsel, the rules and principles discussed below apply with equal force to their practice.

I. Duty of Competence (Rule 1.1)

Since 2021, California’s Rule 1.1 duty of competence has included “the duty to keep abreast of the changes in the law and its practice, including the benefits and risks associated with relevant technology.”ⁱⁱ Recent guidance has since clarified that, when using AI, an attorney’s duty does not end with a one-time assessment of the utility (or reliability) of the tool, but rather requires ongoing monitoring and verification of its outputs. Last year, in People v. Alvarez, the appellate court held that verification of citations is essentially the “responsibility of a competent attorney” and “cannot [be] delegate[d] . . . to any form of technology.”ⁱⁱⁱ COPRAC’s proposed Rule 1.1 amendment, approved for comment March 13, 2026, would similarly require a lawyer, when using technology, to “independently review, verify, and exercise professional judgment regarding any output generated by the technology that is used in connection with representing a client.”^iv This requirement of course applies to trusts and estates practice use cases, such as drafting an instrument (and ensuring the fit and correctness of its terms), preparing a fiduciary accounting (and ensuring the accuracy of its data), and researching a point of probate or other law (and verifying accurate citation).

On May 14, 2026, the Board of Trustees reiterated the importance of attorney review and verification of AI systems in its updated Practical Guidance, the first specific guidance on agentic AI, which has the ability to complete — and even initiate tasks — without human prompting, review, or approval. The 2026 Practical Guidance states: “Lawyers must not deploy agentic systems in a manner that allow the system to make substantive legal determinations, communicate legal advice, prepare and file pleadings, or otherwise act in a representative capacity without meaningful lawyer supervision and review.”^v Thus, the Practical Guidance underscores the importance for an attorney to understand “the nature and extent of an agentic system’s autonomy” and “the safeguards necessary” to prevent ethical violations.^vi

II. Candor to the Tribunal (Rule 3.3) and Meritorious Claims & Contentions (Rule 3.1)

Rule 3.3 prohibits an attorney from knowingly^vii making false statements of fact or law to a tribunal, failing to disclose controlling adverse authority known to the lawyer, or offering evidence the lawyer knows to be false.^viii Although Rule 3.3 prohibits knowing misconduct, not inadvertent error, case law — and proposed rules — make lawyers using AI responsible for AI-produced errors, even if inadvertent, by imposing a specific verification duty on AI-assisted court submissions.

Last year, the court in Shayan v. Shakib held that a brief’s signatory attorney owns its content, regardless of whether AI or another process produced the errors.^ix In part relying on Alvarez, the court in Shakib reasoned that culpable knowledge may be inferred where the AI-tool used to generate citations (or the process used to verify the accuracy of AI-generated citations) involves an inherent risk that the tools and verification process will produce inaccurate statements of law.^x These cases apparently applied a constructive knowledge standard, imputing knowledge of the unreliability of AI tools, apart from what the offending attorney really knew on the subject.

Proposed AI rules dispense with a culpable knowledge limitation, imposing an affirmative duty of verification in using AI tools. COPRAC’s proposed Rule 3.3 comment states that candor “includes the obligation to verify the accuracy and existence of cited authorities, including ensuring no cited authority is fabricated, misstated, or taken out of context, before submission to a tribunal,” applied expressly to “any cited authorities generated or assisted by artificial intelligence.”^xi COPRAC’s March 19, 2026 advisory on AI hallucinations similarly provides that a lawyer who submits AI-generated work containing fabricated authorities remains responsible for it; not knowing of the hallucination is no excuse, because a lawyer must independently confirm citations and key facts before filing court papers.^xii

The Legislature may soon codify these emerging standards. Senate Bill 574, unanimously passed by the Senate 39-0 on January 29, 2026, and as of June 22, 2026, amended and re-referred to the Assembly Committee on Judiciary, would add a verification duty to the Business and Professions Code and amend Code of Civil Procedure section 128.7 to bar a filing containing a citation the responsible attorney has not personally read and verified.^xiii

The foregoing standards similarly apply to the court and its staff, as made clear in additions to the California Standards of Judicial Administration and Rules of Court, effective September 1, 2025, requiring a judicial officer to take reasonable steps to verify that generative AI material is accurate and correct any hallucinated output.^xiv

The 2023 and 2026 Practical Guidance both cite the related Rule 3.1 duty to present meritorious claims and contentions to the tribunal, with the 2026 Practical Guidance noting that use of AI does “not diminish the lawyer’s personal responsibility for the … legal sufficiency of any submission to the court.”^xv

III. Duty of Confidentiality (Bus. & Prof. Code, § 6068(e); Rules 1.6, 1.8.2)

The 2026 Practical Guidance provides, as a general principle, that “a lawyer must not input any confidential information of [a] client into a generative AI solution that may present material risks to confidentiality or security, absent informed client consent . . . as to the underlying risks.”^xvi Thus, in the absence of informed client consent, a lawyer must not input confidential client information into a generative AI tool without first confirming the platform only retains, uses, and shares it in a manner consistent with confidentiality.

Consistent with this principle, Senate Bill 574 would add Business and Professions Code section 6068.1, requiring attorneys using AI to ensure that “confidential, personal identifying, and other nonpublic information is not entered into a public generative artificial intelligence system.”^xvii Likewise, recent additions to the California Standards of Judicial Administration and Rules of Court, “prohibit the entry of confidential, personal identifying, or other nonpublic information into a public generative AI system” by court personnel.^xviii

In evaluating the underlying risks to confidentiality posed by a particular AI platform, the 2026 Practical Guidance directs counsel to assess a tool’s security protections, privacy policy, and to consider consulting an IT or cybersecurity professional before adopting an unfamiliar platform.^xix COPRAC’s proposed additions to the comments of Rule 1.6 would clarify that revealing a client’s confidential information includes exposing such information to technological systems, including AI tools, “where such exposure creates a material risk that the information may be accessed, retained, or used, whether by the technological system or another use of that technological system.”^xx A recent U.S. District Court opinion further highlights the potential pitfalls (for clients and attorneys alike) of consumer AI tools lacking enterprise platform security assurances in their privacy policy, finding that communications with such a tool are not protected by attorney-client privilege or the work-product doctrine, and any privilege that might otherwise attach was waived by that disclosure to the AI provider as a third party.^xxi

IV. Client Authority and Communication (Rules 1.2, 1.4)

Under Rule 1.2 (“Scope of Representation and Allocation of Authority”), a lawyer must follow applicable client instructions or guidelines, which would presumably include any express restrictions or limitations on the use of AI.

The 2026 Practical Guidance further suggests that, under both Rule 1.2 and Rule 1.4 (“Communication with Clients”), lawyers should consider disclosing any intended use of AI in the representation, particularly where use of AI will materially affect the decision-making process.^xxii

COPRAC’s proposed Rule 1.4 comment would convert this disclosure suggestion into a rule, requiring a lawyer to disclose to the client any intended use of AI in the course of the representation where such use of AI “presents a significant risk or materially affects the scope, cost, manner, or decision-making process” of the representation, judged against the technology’s novelty, the risks involved, and the client’s sophistication.^xxiii

V. Fees and Billing (Rule 1.5; Bus. & Prof. Code, §§ 6147-6148)

No proposed rule change addresses billing per se. The 2026 Practical Guidance does, however, state that a lawyer may spend billable time using generative AI to create work product (e.g., crafting or refining AI inputs and prompts, or reviewing and editing AI outputs), but time charges must reflect the time actually spent.^xxiv The 2026 Guidance expressly recognizes that AI may reduce time required for tasks.^xxv On the other hand, at some point time spent on a task may become excessive (regardless of the tool or process used), and billing for excessive time spent correcting or revising AI output — like such time on drafts prepared by other lawyers or paralegals — risks improperly charging an unconscionable fee, contrary to Rule 1.5.^xxvi

Lawyers may consider alternative billing structures, such as flat fees, and building the time savings into their competitive rates for such tasks.^xxvii Fee agreements should state how fees and costs are calculated and the extent to which generative AI factors into the lawyer’s rates. Generally, generative AI subscriptions used to provide general office functionality (e.g., drafting assistance, research capabilities, and document review) should be treated as an overhead expense of the law office and should not be charged separately to clients.^xxviii Lawyers may, however, charge clients for AI costs specifically incurred for their matter, beyond general office operations, such as per use fees for specialized AI tools or costs for custom AI tools developed for a particular matter.^xxix The fee agreement must clearly disclose that AI costs may be billed separately, the charges must reasonably reflect the actual cost, and no markup may be added without the client’s informed written consent.^xxx

VI. Supervisory Duties (Rules 5.1, 5.2, 5.3)

Managing and supervising lawyers must adopt clear AI policies and train staff on its risks. COPRAC’s proposed amendments to Rules 5.1 and 5.3 would require managerial lawyers to establish internal AI policies and extend supervision of nonlawyer assistants to their use of AI tools.^xxxi The Judicial Council’s rule 10.430 imposes a parallel duty on the courts, requiring them to adopt a written AI use policy or prohibit the technology outright.^xxxii

Conclusion

California’s AI regulations have developed considerably in a year, converging on the same verification requirement, and extending AI-specific guidance to well-established rules. For probate practice, no less than other practice areas, the increasing power and efficiency of ever-advancing AI tools must be tempered by adherence to the familiar duties governing attorneys, as adapted and updated to address the promise and pitfalls of navigating this new, exciting frontier.

Download the original publication (PDF) →