Connecticut mandates impact assessments for high-risk artificial intelligence systems by 2026

On April 9, 2025, the Connecticut State Legislature introduced Senate Bill 2, a significant piece of legislation aimed at regulating high-risk artificial intelligence (AI) systems. The bill seeks to establish a framework for the deployment and management of these systems, addressing concerns over their potential risks and impacts on society.

The primary purpose of Senate Bill 2 is to ensure that deployers of high-risk AI systems conduct thorough impact assessments before and during the deployment of such technologies. Key provisions of the bill require that, starting October 1, 2026, any deployer or their contracted third parties must complete an impact assessment that evaluates the intended uses, benefits, and potential risks associated with the AI systems. This includes an analysis of algorithmic discrimination risks and the measures taken to mitigate them.

Notably, the bill mandates that these assessments be conducted annually and within ninety days of any substantial modifications to the AI systems. This proactive approach aims to enhance accountability and transparency in the deployment of AI technologies, which have been increasingly scrutinized for their ethical implications.

Debates surrounding Senate Bill 2 have highlighted concerns from various stakeholders. Proponents argue that the legislation is essential for safeguarding against algorithmic bias and ensuring that AI systems are deployed responsibly. Critics, however, express concerns about the potential burden on businesses and the feasibility of compliance, particularly for smaller companies that may lack the resources to conduct comprehensive assessments.

The implications of Senate Bill 2 extend beyond regulatory compliance; they touch on broader economic and social issues. By establishing a clear framework for AI deployment, the bill could foster public trust in technology, potentially leading to increased adoption and innovation in the sector. Conversely, if perceived as overly restrictive, it could stifle technological advancement and economic growth.

As the bill progresses through the legislative process, experts anticipate that it will spark further discussions on the balance between innovation and regulation in the rapidly evolving field of artificial intelligence. The outcomes of these discussions will likely shape the future landscape of AI governance in Connecticut and potentially serve as a model for other states considering similar legislation.