On January 9, 2024, House Bill H.711, for an act relating to creating oversight and liability standards for developers and deployers of inherently dangerous artificial intelligence systems, was read for the first time in the Vermont House of Representatives and on the same date, referred to the Committee on Commerce and Economic Development. 

Definitions

In particular, the bill defines important terms including, 'artificial intelligence systems,' 'consequential decision,' 'deployer,' 'developer,' 'dual use foundational model,' 'generative artificial intelligence system,' 'high-risk artificial intelligence system,' and 'inherently dangerous artificial intelligence systems.'

More specifically, the bill defines an inherently dangerous artificial intelligence (AI) system as a high-risk AI system, dual-use foundational model, or generative AI system. Further, the bill defines high-risk AI systems as any AI system, regardless of the number of parameters and supervision structure, that is:

• used, reasonably foreseeable as being used, or is a controlling factor in making a consequential decision;
• used, or reasonably foreseeable as being used, to categorize groups of persons by sensitive and protected characteristics, such as race, ethnic origin, or religious belief;
• used, or reasonably foreseeable as being used, in the direct management or operation of critical infrastructure;
• used, or reasonably foreseeable as being used, in vehicles, medical devices, or in the safety system of a product;
• used, or reasonably foreseeable as being used, to influence elections or voters; or
• used to collect the biometric data of an individual from a biometric identification system without consent.

What are the obligations under the bill?

The bill would require each deployer of an AI system to, among other things:

• submit to the Division of Artificial Intelligence (the Division), an AI system safety and impact assessment, prior to deploying the inherently dangerous AI system in Vermont, and every two years thereafter; and
• submit to the Division an updated AI system safety and impact assessment if the deployer makes a material and substantial change to the inherently dangerous AI system that includes:
  • the purpose for which the system is used for; or
  • the type of data the system processes or uses for training purposes.

Additionally, the bill would impose a duty of care on deployers and developers of inherently dangerous AI systems that could be reasonably expected to impact consumers, requiring them to exercise reasonable care to avoid any reasonably foreseeable risk arising out of the development, intentional and substantial modification, or deployment of an AI system that causes or is likely to cause the acts as listed in the bill.

What are the prohibitions under the bill?

The bill also lays down certain prohibitions on unsafe AI products, including, among others, that no developer shall offer, sell, lease, give, or otherwise place in the stream of commerce:

• an inherently dangerous AI system, unless the developer has conducted documented testing, evaluation, verification, and validation of that system at least as stringent as the latest version of the Artificial Intelligence Risk Management Framework published by the National Institute of Standards and Technology (NIST); or
• an AI system that creates reasonably foreseeable risks, unless the developer mitigates these risks to the extent possible, considers alternatives, and discloses vulnerabilities and mitigation tactics to a deployer.

Enforcement

Notably, the bill would provide for a private right of action, allowing consumers harmed by a violation of the bill or rules adopted thereunder to bring an action in the Superior Court for damages incurred, injunctive relief, punitive damages in the case of an intentional violation, and reasonable costs and attorney's fees. Additionally, the Attorney General of Vermont would be empowered to enforce the provisions of the bill and may bring an action against a deployer or developer for noncompliance, to restrain by temporary or permanent injunction the noncompliance.

If enacted, the bill would enter into effect on July 1, 2024.

You can read the bill here and track its progress here.