California’s New AI Law: What SB 574 Means for Clients, Litigation and Arbitration

Artificial intelligence is rapidly changing how attorneys research legal issues, draft documents, review evidence, and communicate with clients. Used properly, it can reduce time and expense. Used carelessly, it can invent legal authorities, produce incorrect advice or expose confidential client information. California has now made it clear that lawyers, not their software, are responsible for the results.

On September 30, 2026, Governor Gavin Newsom signed Senate Bill 574 (SB 574), Chapter 858 of the Statutes of 2026. Effective January 1, 2027, the law establishes express statutory duties for California lawyers using generative artificial intelligence (AI), revises rules governing court filings, and restricts arbitrators’ reliance on AI. It permits the responsible use of AI; it does not permit lawyers to outsource their judgment. [1]

A Lawyer Cannot Delegate the Practice of Law to AI

New Business and Professions Code section 6068.1 states that an attorney may not delegate the practice of law to generative AI. That does not prohibit a lawyer from using AI to organize records, generate a draft, identify research questions or summarize testimony. It means the attorney must make the legal decisions, evaluate the evidence and take professional responsibility for advice and work product. The statute expressly preserves existing duties of competence and diligence. [1]

For clients, that distinction is critical. An AI-generated contract provision or litigation strategy may look persuasive but it could overlook a controlling statute, a contractual exception or a material fact. The lawyer must determine whether it is actually correct and appropriate for the client’s objectives.

Client Confidentiality: A New Statutory Safeguard

Section 6068.1 restricts an attorney from entering confidential, personally identifying or other nonpublic information into an AI system unless access to that information is limited to the attorney and persons authorized by the attorney who are obligated to protect its confidentiality. The statute specifically covers financial information, account numbers, dates of birth, medical information and other sensitive identifiers. [1]

This provision is especially relevant to financial advisers, broker-dealers, businesses and individuals involved in investigations or litigation. Brokerage records, customer data, trading information, acquisition plans and privileged communications cannot simply be pasted into an unprotected public chatbot. A paid subscription by itself is not proof of compliance. Lawyers must understand a provider’s access controls, confidentiality commitments and data-handling practices before uploading sensitive material.

Accuracy Is Mandatory and Every Court Citation Must Be Checked

Lawyers who use AI must take reasonable steps to verify its output, including legal and statutory citations, and correct erroneous or fabricated material they use. Separately, SB 574 amends Code of Civil Procedure section 128.7 to prohibit citations in briefs, motions, pleadings, or other court papers unless an attorney responsible for the filing has personally verified them. Significantly, this obligation applies to all citations, not merely those produced by AI, and extends to pro per parties as well. [1]

The risks are not hypothetical. In Noland v. Land of the Free, a California appellate court sanctioned an attorney $10,000 after briefs contained fabricated case quotations and citations produced by AI. The court also directed that the opinion be sent to the State Bar. SB 574 gives that warning additional statutory force. [2]

Under CCP 128.7, violations can lead to court-imposed sanctions and payment of the opposing parties’ attorney’s fees. Clients are entitled to expect that cases cited on their behalf exist, say what the lawyer claims they say, and have been independently checked.

AI Use in Court Must Be Disclosed

Another significant change is the requirement that attorneys disclose their use of generative AI to the court for documents submitted to it. The law does not specify a uniform disclosure form or exactly how much AI involvement triggers that requirement; individual court rules and orders may impose additional directions. Lawyers should therefore establish a consistent disclosure procedure rather than assume that AI-assisted drafting can remain invisible. [1]

For public-facing content, including website articles, client alerts, and marketing materials, the statute requires attorneys to consider whether to disclose AI use. It does not impose the same disclosure requirement for every public communication. Separately, the State Bar’s practical guidance calls for thoughtful client communication about the benefits and risks of AI use. [3]

Arbitration Decisions Must Remain Human Decisions

SB 574 also adds Code of Civil Procedure section 1282.1. Arbitrators may not delegate any part of their decision-making process to generative AI. Nor may they rely on AI-generated information outside the arbitration record without advance disclosure to the parties and, as far as practicable, an opportunity for the parties to comment. This is particularly important in securities, employment, and commercial arbitration, where outcomes often turn on disputed testimony and complex financial records. [1]

Other provisions require the Judicial Council to revisit its existing AI standards for judges. They also revise complaint procedures under the State Bar’s voluntary certification program for alternative dispute resolution providers. Those provisions are distinct from the duties imposed directly on attorneys and arbitrators. [1, 4]

What Clients Should Expect From Their Law Firm

SB 574 does not mean that clients should avoid firms that use AI. Quite the opposite: when carefully supervised, technology can expedite document review, sharpen issue-spotting, and make legal services more efficient. The relevant questions are whether a firm protects confidential data, independently verifies its work, and keeps attorneys accountable for substantive decisions.

Clients may also reasonably ask how AI-related efficiencies are reflected in their bills. The State Bar’s guidance explains that an attorney may charge for actual time spent prompting, reviewing, and editing AI-assisted work, but not for hours supposedly worked that AI saved. Engagement terms should accurately explain any separately charged AI-related costs. [3]

For a business facing a regulatory investigation, a financial professional in a FINRA dispute, or a party to significant commercial litigation, these protections matter. AI should help counsel work faster and more effectively, not compromise privileged information, create unreliable evidence, or substitute machine-generated answers for legal judgment.

The central message of SB 574 is straightforward, technology may assist the lawyer, but responsibility to the client remains with the lawyer. That is a sound principle, and one clients should insist upon.

Shustak Reynolds & Partners, P.C. focuses its practice on securities and financial services law and complex business disputes.
We represent advisors, financial professionals, broker-dealers, registered representatives, investors and businesses.
Attorney Erwin J. Shustak can be reached in the firm’s San Diego office at (619) 696-9500.

[1] SB 574 (2025–2026 Reg. Sess.), ch. 858, Stats. 2026; Bus. & Prof. Code §§ 6068.1, 6173; Code Civ. Proc. §§ 128.7, 180, 1282.1. https://leginfo.legislature.ca.gov/faces/billTextClient.xhtml?bill_id=202520260SB574

[2] Noland v. Land of the Free, L.P. (2025) 114 Cal.App.5th 426. https://law.justia.com/cases/california/court-of-appeal/2025/b331918.html

[3] State Bar of California, Practical Guidance for the Use of Generative Artificial Intelligence in the Practice of Law (updated May 14, 2026). https://www.calbar.ca.gov/Portals/0/documents/ethics/Generative-AI-Practical-Guidance.pdf

[4] California Standards of Judicial Administration, standard 10.80 (effective Sept. 1, 2025). https://courts.ca.gov/cms/rules/index/standards/Standard10_80