Newsom Signs California Laws Requiring Human Review of AI Firing Decisions

California Enacts AI Rules Requiring Lawyers to Check Citations and Disclose Use

California is keeping lawyers responsible for legal work, not barring every use of AI. Gov. Gavin Newsom signed SB 574 on September 30, 2026. The measure requires human verification of court citations and disclosure of AI-created filings, while restricting confidential data entry and prohibiting lawyers from handing the practice of law to AI.

Bloomberg Law describes the measure as the first state legislation regulating attorneys’ use of generative AI. It codifies and extends existing state court rules. California’s courts newsroom says the requirement to disclose AI use in documents submitted to court begins in 2027.

The verification requirement is broader than a check for AI mistakes. An attorney responsible for submitting a court document must personally verify every citation it contains, including those supplied by generative AI. That obligation covers briefs, pleadings, motions and other papers filed in court—not only documents known to contain machine-generated references.

The law also requires reasonable steps to verify AI outputs and correct erroneous or hallucinated material—content the system produces that is false or invented. These are distinct duties: reviewing generated work for accuracy does not replace the personal citation check required for a filing.

At least one attorney must sign each court filing in their individual name, with an exception for people representing themselves without a lawyer. Bloomberg Law places these requirements against a recurring problem in courts: AI-generated citations that are invented or badly interpret prior case law.

A brief, pleading, motion, or any other paper filed in any court shall not contain any citations that an attorney responsible for submitting the pleading has not personally verified, including any citation provided by generative artificial intelligence.

SB 574 bars lawyers from entering confidential, personally identifying or other nonpublic information into a generative AI system unless access is restricted. The permitted users must be the attorney and authorized people who are themselves bound by confidentiality obligations.

The distinction is between open-access tools and systems with restricted access, rather than a blanket prohibition on using AI with sensitive information. The exception ties permission to who can access the information and their duty to protect it.

Disclosure obligations also depend on where the work goes. Esquire Deposition Solutions’ October 6 explanation on JD Supra distinguishes a mandatory court-filing rule from a more qualified duty for public content:

Court documents: Attorneys must disclose when generative AI was used to create a document submitted to court.

Content shared with the public: Lawyers must consider whether to disclose generative AI use; that is not the same as mandatory disclosure in every instance.

The law reaches beyond attorneys’ filings. Arbitrators are barred from delegating any part of their decision-making to generative AI. They must also disclose to the parties when they rely on AI-generated information. Disclosure does not create an exception to the separate ban on delegating decisions.

SB 574 sets parameters for courts to impose sanctions for violations. Esquire’s explanation says AI-related sanctions disputes should initially be worked out among attorneys before reaching the court through a public-docket filing. It interprets that procedure as addressing the reputational harm that can accompany discovery of AI errors in court documents.

California Requires Employers to Identify AI-Driven Layoffs in Covered Notices

Starting January 1, 2027, covered California employers must identify when AI or other automation substantially drives a mass layoff, relocation or termination. Their notices must describe the affected jobs, tasks and technology. Gov. Gavin Newsom signed SB 951 on September 30, 2026, adding those obligations to the state’s existing workforce-notice framework.

The law adds a technology-specific layer to Cal/WARN, California’s rules governing notices for covered workforce reductions. The trigger is causation: AI or other automated technology must cause the displacement in whole or in substantial part. The disclosure requirement therefore extends beyond AI to other technology that replaces positions or automates their work.

Sponsored by Senator Eloise Gómez Reyes, the measure concerns technological-displacement notices. It amends three sections of California’s Labor Code—1400.5, 1401 and 1402.7—within the existing Cal/WARN framework.

CDF Labor Law’s explanation says the definition of a covered establishment remains unchanged. It covers industrial or commercial facilities with 75 or more employees currently, or within the previous 12 months. That boundary matters: SB 951 adds information to notices within the Cal/WARN framework, rather than creating a disclosure requirement for every workplace using automation.

The framework’s employer definition includes a person who directly or indirectly owns and operates a covered establishment. It also includes a parent corporation when a subsidiary directly owns and operates that establishment. An employee is someone employed for at least six months of the 12 months before the date notice is required.

The required header is only the start. Employers must supply three sets of details, connecting the workforce reduction to both the work being automated and the technology they will use. A notice must include:

The number of automation-related layoffs, the affected job classifications or occupations, and the locations where those employees work.

The job functions that will be automated, identifying the work that technology will take over.

The specific category or type of AI or other automated technology the employer will use.