Alaska SB 177, Section 2. ("Use by State Agencies of Artificial Intelligence")

Proposed 2024-01-16 | Official source

Summary

Requires departments to inventory AI systems in state agencies biannually, publish findings online, and assess impacts. Prohibits using AI for biometric identification, emotion recognition, manipulation, or social scoring. Requires agency consent for sensitive data use.

  • This machine-generated summary is awaiting review by an AGORA editor. Use with caution.

Key facts

🏛️ This document was proposed and/or enacted by the State of Alaska but is now defunct. For authoritative text and metadata, visit the official source.

📜 This document's name is Alaska Senate Bill No. 177. AGORA also tracks this document under the name Alaska SB 177, Section 2. ("Use by State Agencies of Artificial Intelligence"). It is part of Alaska SB 177.

↳ This document is part of a longer one: Alaska SB 177. Some AGORA documents are "split off" from longer documents that mix AI and non-AI content, such as omnibus authorization or appropriations laws in the United States Congress. Read more >>

Themes AI risks, applications, governance strategies, and other themes addressed in AGORA documents.

Thematic tags are in progress.

Full text

  • This is an unofficial copy. The document has been archived and reformatted in plaintext for AGORA. Footnotes, tables, and similar material may be omitted. For the official text, visit the original source.
Sec. 2. AS 44.99 is amended by adding new sections to read: Article 7. Use by State Agencies of Artificial Intelligence and Data about Individuals. Sec. 44.99.700. Inventory. (a) Every two years, the department shall conduct an inventory of all systems used by state agencies that employ artificial intelligence for consequential decisions. Each state agency shall assist the department as necessary. An inventory must include, at a minimum, the following information for each system: (1) the name of the system; (2) the vendor that provides the system, if any; (3) a description of the general capabilities and uses of the system; and (4) whether the state agency completed an impact assessment of the system under AS 44.99.710 before the system's implementation. (b) Upon completion, the department shall publish each inventory on the department's Internet website. Sec. 44.99.710. Impact assessments. (a) At least once every two years, the head of a state agency that uses a system that employs artificial intelligence for consequential decisions shall conduct an impact assessment of the system. An impact assessment must include, at a minimum, an analysis of (1) the efficacy of the system; (2) the human oversight involved in the system; (3) the accountability mechanisms in place for the system; (4) the process by which an individual may appeal a decision made or facilitated by the system; (5) the current and potential benefits, liability, and risks to the state from the system, including risks related to cybersecurity and intellectual property and any measures used to mitigate liability and risks; (6) the current and potential effects of the system on the liberty, finances, livelihood, and privacy interests of individuals in the state, including effects from any use of geolocation data by the system; (7) any unlawful discrimination against or unlawful disparate impact on an individual or a group of individuals that has resulted or may result from the system; and (8) the policies and procedures that govern the process of using the system for consequential decisions. (b) Upon completion, the state agency that conducts the impact assessment shall provide the assessment to the department. Upon receiving an assessment, the department shall publish the assessment on the department's Internet website. Sec. 44.99.720. Requirements for use of artificial intelligence by state agencies. (a) A state agency that uses a system that employs artificial intelligence for consequential decisions shall (1) notify each individual who may be legally or significantly affected by the use of the system; (2) obtain an individual's consent before soliciting or acquiring sensitive personal data about the individual that will be used by the system; (3) provide an appeals process that includes manual human review for an individual who is legally or significantly affected by the use of the system; and (4) inform a prospective employee of the state agency about any video interview that involves the use of artificial intelligence and obtain the prospective employee's consent before employing artificial intelligence. (b) A state agency may not use a system that employs artificial intelligence for consequential decisions if the system involves (1) biometric identification, including facial recognition; (2) emotion recognition; (3) cognitive behavioral manipulation of individuals or groups; or (4) social scoring. (c) A state agency may not use a system that employs artificial intelligence for consequential decisions if the system uses data hosted in (1) the People's Republic of China, including the Hong Kong Special Administrative Region and Macao Special Administrative Region; (2) the Republic of Cuba; (3) the Islamic Republic of Iran; (4) the Democratic People's Republic of Korea; (5) the Russian Federation; or (6) the Bolivarian Republic of Venezuela under the regime of Nicolás Maduro Moros. (d) A state agency may contract with a person for a system that employs artificial intelligence for consequential decisions only if the person has implemented multi-factor authentication to secure the system and data stored by the system. Sec. 44.99.730. Transfer of data between state agencies. Unless required by law, a state agency may not transfer data about an individual to another state agency without the individual's consent. Sec. 44.99.740. Regulations. (a) The department shall adopt regulations under AS 44.62 (Administrative Procedure Act) concerning the development, procurement, implementation, use, and ongoing assessment of systems that employ artificial intelligence by state agencies for consequential decisions. The regulations must include, at a minimum, provisions that (1) govern the procurement, implementation, and ongoing assessment of each system; (2) require a state agency to conduct an impact assessment of each system under AS 44.99.710 before its implementation; (3) ensure that a system does not result in unlawful discrimination or an unlawful disparate impact on an individual or a group of individuals; and (4) provide for the ongoing assessment of each system. (b) The department may adopt additional regulations under AS 44.62 (Administrative Procedure Act) necessary to implement AS 44.99.700 - 44.99.730. Sec. 44.99.750. Civil liability for harm. (a) An individual who suffers harm as a result of a violation of AS 44.99.700 - 44.99.730, a violation of a regulation adopted under AS 44.99.740, or gross negligence or reckless or intentional misconduct relating to the use of artificial intelligence by a state agency or state employee may bring a civil action in the superior court against the state or state employee. (b) An individual who suffers harm under (a) of this section may recover damages for the harm to the individual, punitive damages under AS 09.17.020, and full reasonable attorney fees and costs in a civil action brought under this section. Sec. 44.99.760. Definitions. In AS 44.99.700 - 44.99.760, (1) "artificial intelligence" means an automated system that uses data input, human-defined objectives, and machine learning, natural language processing, or other computational processing techniques of similar or greater complexity to make a decision or facilitate human decision making; (2) "biometric identification" means the analysis of an individual's physical or behavioral characteristics to uniquely identify the individual; (3) "cognitive behavioral manipulation" means the use of a subliminal technique for the purpose of influencing an individual's behavior to achieve a desired outcome; (4) "consequential decision" means a conclusion, decision, or judgment by a state agency that can have a legal or significant effect on an individual; (5) "department" means the Department of Administration; (6) "emotion recognition" means the analysis of an individual's bodily expressions, including facial and verbal expressions, to identify or predict the individual's emotions; (7) "individual" means a natural person; (8) "sensitive personal data" means (A) data that reveals an individual's racial or ethnic origin, political opinions, or religious or philosophical beliefs; (B) an individual's genetic data; (C) an individual's biometric data when used for biometric identification; or (D) an individual's geolocation data; (9) "social scoring" means evaluating, classifying, rating, or scoring the trustworthiness or social standing of an individual based on behavior or socioeconomic, political, or religious status; (10) "state agency" means the University of Alaska, a public corporation of the state, or a department, institution, board, commission, division, authority, committee, or other administrative unit of the executive branch of state government.