Regulates the development and use of certain artificial intelligence systems to prevent algorithmic discrimination; requires independent audits of high risk AI systems; provides for enforcement by the attorney general.
This bill, the "Responsible AI Safety and Education Act" (RAISE Act), regulates large artificial intelligence (AI) developers who train or use "frontier models" - defined as models requiring over $100 million in computing costs or 10^26 computational operations. It directly affects commercial AI companies (excluding academic researchers) by requiring them to implement safety protocols, conduct risk testing, and report incidents like unauthorized model access or misuse. Key provisions mandate documented safety measures to prevent "critical harm" (defined as serious injury to 100+ people or $1 billion+ in property damage), with remedies for violations. The law also creates specific definitions for terms like "frontier model," "safety incident," and "knowledge distillation" (modified AI models), ensuring clarity for compliance.
Establishes the artificial intelligence training data transparency act requiring developers of generative artificial intelligence models or services to post on the developer's website information regarding the data used by the developer to train the generative artificial intelligence model or service, including a high-level summary of the datasets used in the development of such system or service.
Requires state units to purchase a product or service that is or contains an algorithmic decision system that adheres to responsible artificial intelligence standards; specifies content included in responsible artificial intelligence standards; requires the commissioner of taxation and finance to adopt certain regulations; alters the definition of unlawful discriminatory practice to include acts performed through algorithmic decision systems.
Regulates the development and use of certain artificial intelligence systems to prevent algorithmic discrimination; requires independent audits of high risk AI systems; provides for enforcement by the attorney general.
Establishes the position of chief artificial intelligence officer and such person's functions, powers and duties; including, but not limited to, developing statewide artificial intelligence policies and governance, coordinating the activities of any and all state departments, boards, commissions, agencies and authorities performing any functions using artificial intelligence tools; makes related provisions.
Directs the commissioner of education to make recommendations to the board of regents regarding the incorporation of instruction in artificial intelligence system literacy into the school curriculum at the elementary, junior high and senior high levels; directs the board of regents to adopt guidelines if the curriculum is adopted.
Requires that every newspaper, magazine or other publication printed or electronically published in this state, which contains the use of generative artificial intelligence or other information communication technology, shall identify that certain parts of such newspaper, magazine, or publication were composed through the use of artificial intelligence or other information communication technology.
Requires the owner, licensee or operator of a visual or audio generative artificial intelligence system to take steps to prohibit its users from creating unauthorized realistic depictions of public officials.
The RAISE Act establishes safety requirements for developers of powerful artificial intelligence models, known as "frontier models." A frontier model is defined as one trained with over 10^26 computational operations (costing more than $100 million) or a distilled version costing over $5 million. Large developers must implement documented safety protocols to prevent "critical harm," defined as events causing death, serious injury to 100+ people, or $1 billion in damages from AI misuse. The bill also defines key terms like "safety incident" and outlines remedies for violations, focusing on risk mitigation rather than restricting AI development.
