What California Legal Professionals Using Generative Artificial Intelligence (AI) Need to Know About California’s SB 574

On September 1, 2026, the California legislature approved Senate Bill 574 (SB 574), a first-of-its-kind state measure that creates a statutory framework governing how attorneys and arbitrators use generative AI. The landmark bill, introduced by Senator Tom Umberg, transforms existing advisory ethics guidelines for California lawyers into binding legal requirements. SB 574 is expected to fundamentally shift daily workflows for law firms and legal practitioners across the state, emphasizing personal accountability and client protection. The bill awaits signature from Governor Gavin Newsom.

To keep up with the latest in Artificial Intelligence (AI) use and trends in law firms, see our CLE webinars. CLE credit is available in California and numerous jurisdictions.

New Statutory Duties for Attorneys

While legal professionals are already bound by duties of competence and confidentiality, SB 574 codifies these obligations directly in the context of generative AI use. The bill introduces critical mandates for practitioners across several key areas:

• Strict Confidentiality Protections: Attorneys are strictly prohibited from entering confidential, personally identifying, or non-public client information into public generative AI systems. This requirement directly addresses the risk of AI platforms utilizing client input data for model training. Moving forward, practitioners cannot simply paste client documents into publicly available chatbots without risking a breach of their statutory duty. This is usually an issue when using free, consumer-oriented AI tools rather than purpose-built, legal-specific tools.

• Verification and Anti-Discrimination: The bill requires attorneys to take “reasonable steps” to verify the accuracy of all AI-generated material and to correct any erroneous or “hallucinated” outputs. Furthermore, attorneys must ensure that their use of generative AI tools does not unlawfully discriminate against or disparately impact protected classes. The verification requirement is really not much different than the requirement that attorneys have been taught for generations to confirm that any cases they cite are still good law (see below).

• The Citation Review Mandate: Perhaps the most disruptive element of SB 574 is its amendment to the California Code of Civil Procedure Section 128.7. The bill mandates that the attorney responsible for submitting a court filing must personally read and verify every single citation it contains, including those provided by generative AI. This means that the signing attorney—often a senior partner—must be extra diligent when delegating cite-checking tasks to junior associates or paralegals. The rule effectively eliminates the “delegation defense” for hallucinated case law, holding the signing attorney personally responsible for the accuracy of the record.

Preserving Human Judgment in Arbitration

SB 574 also introduces new guardrails for alternative dispute resolution by establishing the new California Code of Civil Procedure Section 1282.1. Recognizing the risk of automated justice, the bill explicitly prohibits arbitrators from delegating any part of their decision-making process to generative AI. AI tools cannot replace an arbitrator’s independent analysis of the facts, evidence, and applicable law.

Additionally, if an arbitrator contemplates using AI-generated information that falls outside the official record, they must provide advance disclosure to all parties and offer an opportunity for comment. Ultimately, the arbitrator must assume full responsibility for all aspects of the final award.

Enforcement and Moving Forward

SB 574 utilizes the existing enforcement infrastructure. Violations of the statute will be addressed through court sanctions and the State Bar disciplinary process. Hundreds of cases nationwide have already involved attorneys submitting briefs with hallucinated citations; SB 574 establishes a clear statutory basis to penalize such negligence.

While SB 574 does not outright ban the use of AI in legal practice, it sends an unmistakable message: technological convenience cannot supersede professional responsibility. Whether or not the Governor signs the bill into law, the principles behind it reflect existing judicial expectations. Legal professionals should proactively audit their firm’s AI policies, invest in legal-specific AI tools to protect client data, and restructure their drafting processes to accommodate mandatory personal citation reviews. In the age of generative AI, the oldest rule of lawyering remains the most important: if the name on the filing is yours, the final responsibility is yours.