S 1169

As amended on 6/9/2025: - Requires developers and deployers to take reasonable care to prevent foreseeable risk of algorithmic discrimination. - Requires a deployer to disclose that AI systems will be used to make a decision and allow the consumer to opt out of the automated process for human review. Requires that after a decision, a deployer must provide an explanation of the process and inform the end user of an opportunity to appeal. - Makes the developer or deployer legally responsible for the quality and accuracy of all consequential decisions made, including any bias, algorithmic discrimination, or misinformation. - Creates whistleblower protections for AI employees. - Requires third-party audits of high-risk AI systems at regular intervals. Requires reporting to the attorney general. Requires implementation of a risk management policy. - Prohibits social scoring. - Creates a private right of action. - The 6/9/2025 amendment adds a definition for "risk management policy and program" and amends the definitions of "deployer" and "substantial change." Exempts systems used for internal purposes and not used in a consequential decision. Amends duty of care to prevent algorightmic bias. Exempts deployers with an urgent necessity to confer a benefit to an end user. Limits end users to one opt-out or one appeal per consequential decision in a six-month period. Expands whistleblower protections to all employees. Amends auditing. Amends reporting requirements. Creates a safe harbor for developers and a rebuttable presumption.