California Senate Bill 420: Automated Decision Systems (State AI Transparency Act)

Applies To

Overview

California SB 420 is a pending state legislative measure introduced in the 2025 legislative session that targets high-risk automated decision systems used in public-facing contexts. The bill would mandate pre-deployment impact assessments to evaluate potential harms, biases, and civil rights implications of covered systems. It establishes transparency reporting requirements, obligating covered entities to disclose how automated decision systems function and affect individuals. Enforcement authority is vested jointly in the California Attorney General and the Civil Rights Department, both of which may bring civil actions against non-compliant organizations. The bill reflects California's broader legislative strategy to regulate AI accountability ahead of any comprehensive federal framework. As of early 2025, the bill remains in the legislative process and has not been enacted into law.

Key Requirements

• •Conduct and document impact assessments for all high-risk automated decision systems prior to public deployment
• •Disclose to affected individuals that automated decision systems are being used to make or assist decisions affecting them
• •Submit AI transparency reports to the relevant California authority on a schedule to be defined in final rulemaking
• •Remediate identified harms or biases surfaced during impact assessments before system deployment proceeds
• •Maintain records of impact assessments and transparency disclosures sufficient to support regulatory review
• •Subject to civil enforcement actions by the California Attorney General or Civil Rights Department for violations, with penalty thresholds not yet finalized in current bill text

What Your Organization Must Do

• →Audit all automated decision systems currently in use or under development to determine which meet the bill's definition of high-risk, creating a documented inventory before the bill advances
• →Establish a standardized impact assessment methodology and workflow that can be applied to high-risk systems ahead of any deployment decision
• →Assign ownership of impact assessment processes to a designated compliance or responsible AI function within the organization
• →Draft or update internal AI transparency disclosure templates to satisfy anticipated individual-notification requirements
• →Update procurement and vendor management contracts to require third-party AI vendors to provide impact assessment documentation for any high-risk system supplied to your organization
• →Monitor California legislative calendar and committee activity on SB 420 to identify the compliance timeline and trigger dates for mandatory assessments