CA AB2930

Requires deployers and developers of automated decision systems to conduct impact assessments before deployment and annually thereafter to evaluate risks of algorithmic discrimination based on protected characteristics.

Mandates deployers notify individuals subject to consequential decisions that an automated decision system is being used and provide plain language descriptions of how the system works and its limitations.

Prohibits deployers from using automated decision systems and developers from making them available if impact assessments identify reasonable risks of algorithmic discrimination until those risks are mitigated.

Grants the Civil Rights Department authority to investigate violations and bring civil actions, with penalties up to $25,000 per violation for algorithmic discrimination cases; provides 45-day cure period before injunctive relief proceedings.

Requires deployers to provide individuals with explanations of consequential decisions, identify principal factors influencing outcomes, and offer opportunities to correct inaccurate personal data; if technically feasible, must accommodate requests for alternative selection processes.