IL SB3590

Establishes product liability framework for AI developers who fail to exercise reasonable care in design, provide adequate warnings, or conform to express warranties for high-impact and generative AI systems.

Defines "high-impact AI systems" as those used in consequential decisions affecting criminal justice, housing, employment, credit, education, healthcare, insurance, critical infrastructure, medical devices, or vehicles.

Holds deployers liable as developers when they make material changes to an AI product or intentionally misuse it contrary to express warranties, with exemptions for businesses with fewer than 20 employees or under 10,000 users.

Creates rebuttable presumption that a product is not defective if the deployer conducted documented testing consistent with NIST AI Risk Management Framework, mitigated foreseeable risks, disclosed risks to consumers, and maintained AI data sheets.

Applies comparative negligence standard allowing joint and several liability for developers and deployers, with plaintiffs' recovery reduced proportionally by their own fault but not barred entirely.