An act relating to creating oversight and liability standards for developers and deployers of inherently dangerous artificial intelligence systems

This bill establishes oversight and liability standards for developers and deployers of inherently dangerous artificial intelligence systems in Vermont. It requires these entities to submit safety and impact assessments to the state before deploying such systems and every two years thereafter, with additional requirements when significant changes are made. The legislation defines inherently dangerous AI systems as those that pose serious risks to public safety, health, or economic security, including systems that can make consequential decisions affecting critical areas like employment, housing, and healthcare. The state Attorney General is authorized to enforce compliance through legal action, which may include injunctions or revocation of business certificates for noncompliance. Small businesses using off-the-shelf AI products according to their terms of use are explicitly excluded from these requirements.