Employers and agencies would need reviews before using AI systems

The bill would regulate employers’ and public agencies’ use of AI systems for worker monitoring and public services. The rules cover workplace monitoring and automated employment decisions, plus automated decisions by public entities about benefits or services. Before using covered workplace systems, employers and public entities would need an impact assessment, and workplace monitoring and decision tools would have to serve listed work purposes and use only necessary data and the least invasive means. Before using an automated system to decide public benefits or services, a public entity would need an assessment by the Department of Labor and Workforce Development. People could not be denied benefits or services based only or decisively on system outputs or monitoring data; a human reviewer must assess relevant information. New systems would need an assessment before use. Assessments for systems already in use when the act takes effect would be due within six months for workplace tools and one year for public benefit or service tools.