Bill H.341 aims to establish safety standards and oversight for developers and deployers of inherently dangerous artificial intelligence (AI) systems in Vermont by introducing a new chapter, Chapter 118, to Title 9 of the Vermont Statutes. The bill emphasizes the responsibilities of developers and deployers to ensure the safety of AI systems that can operate without direct human oversight, particularly in ways that could affect individual rights. Notably, small businesses using off-the-shelf AI products are exempt from these new regulations. The bill mandates that deployers submit an Artificial Intelligence System Safety and Impact Assessment to the Division of Artificial Intelligence prior to deployment and every two years thereafter, detailing the system's purpose, risks, and data processing practices.

Additionally, the bill prohibits developers from offering inherently dangerous AI systems unless they have conducted rigorous testing in accordance with the latest Artificial Intelligence Risk Management Framework from the National Institute of Standards and Technology (NIST). Deployers are required to establish a risk management policy consistent with NIST guidelines. The bill also provides a private right of action for consumers harmed by violations, allowing them to seek damages and legal costs, and creates a rebuttable presumption of compliance for those adhering to the outlined standards. The regulations will apply to entities not classified as small businesses and are set to take effect on July 1, 2025.