AI Frontier Institute

An act relating to the regulation of social media platforms and artificial intelligence systems

View primary source →

Vermont H.365 has not been enacted; it was referred to the Committee on Commerce and Economic Development on 2025-02-26. If enacted, it would require providers of social media platforms (new 9 V.S.A. chapter 118, Sec. 2) and providers of artificial intelligence systems (new 9 V.S.A. chapter 119, Sec. 3) to register annually with the Secretary of State by January 31 and pay a $100 fee. They would give contact information and their current privacy policy and terms. AI providers must also describe the model (capacity, training data, intended use, design process and methodologies) and their data collection, storage and security practices (§ 4194b). Registrants must agree to product safety and privacy terms (§§ 4193c, 4194c) and any terms the Attorney General sets by rule. Failure to register brings a civil penalty of $50 per day, capped at $10,000 per year, plus the unpaid fees and other penalties; the Attorney General may sue in Superior Court (§§ 4193b, 4194b). For social media providers that know or should know a user is a minor, § 4193c would bar design features harmful to minors (including encouraging compulsive use) and dark patterns. It would require high-privacy default settings and bar default contact from unknown adults unless the minor initiates it. It would also prohibit discrimination based on protected characteristics, require reasonable care against heightened risks from processing minors' personal data, limit use of minors' personal identifying information to the purpose it was collected for, and bar targeted advertising to, or sale of data of, minors. AI providers (§ 4194c) must use reasonable care against known or foreseeable algorithmic discrimination, clearly disclose that the consumer is interacting with an AI system, and get informed consent before collecting or using consumer data. They need separate informed consent before sharing or selling that data, and must reasonably secure consumer data used to train the model. Intentional violations would be unfair or deceptive acts under 9 V.S.A. § 2453. The Attorney General may adopt rules (§§ 4193d, 4194d). The act would take effect July 1, 2025 (Sec. 4).

← Back to the tracker