SB2203 creates the Preventing Algorithmic Discrimination Act, a new Illinois law regulating the use of automated decision tools and artificial intelligence in high-stakes decisions. The bill defines covered tools broadly and applies to decisions affecting employment, education, housing, utilities, family planning, healthcare, financial services, criminal justice, legal services, voting, and access to benefits or penalties. It requires deployers to conduct annual impact assessments, assess significant updates, and evaluate potential adverse impacts on protected classes such as race, sex, age, disability, religion, national origin, and others.
The bill also requires advance notice to individuals when an automated decision tool is used in a consequential decision, along with a plain-language description of the tool and contact information for the deployer. In some cases where a decision is made solely by the tool, the bill requires a technically feasible alternative selection process or accommodation if requested. Covered deployers must also maintain a governance program with administrative and technical safeguards, designate an employee responsible for compliance, publish a public policy describing their use of automated decision tools, and submit impact assessments to the Attorney General within 60 days of completion.
SB2203 would add a new standalone act to Illinois law and make conforming changes to the Consumer Fraud and Deceptive Business Practices Act so that violations are treated as unlawful practices enforceable by the Attorney General. It would create new compliance duties for businesses, state agencies, and local government agencies that use automated decision tools in consequential decisions, while exempting smaller deployers in some circumstances. The bill also authorizes civil actions beginning January 1, 2028, and allows administrative fines of up to $10,000 per violation for failure to submit required impact assessments.
No committee transcripts or recorded votes were provided, so there is no documented floor or committee sentiment in the available materials. Based on the bill text alone, the measure appears motivated by concern over algorithmic bias and transparency in AI-driven decision-making, and it is framed as a consumer protection and civil rights measure. The overall tone of the proposal is regulatory and precautionary rather than permissive.
The main points of potential contention are the scope of the regulation, the compliance burden on deployers, and the bill’s private right of action and enforcement provisions. Businesses and public agencies using AI tools may object to annual impact assessments, notice requirements, governance-program mandates, and Attorney General reporting obligations, especially for tools used in employment, housing, healthcare, finance, and criminal justice. Another likely issue is the bill’s broad definition of algorithmic discrimination and the possibility of litigation over whether a tool caused actual harm, while supporters would likely emphasize transparency, accountability, and anti-discrimination protections.