An Act Addressing Innovations In Artificial Intelligence.
SB 1249 is a broad artificial intelligence policy bill that combines state government data-management changes, a proposed AI regulatory sandbox, liability rules for generative AI interactions, and a state investment vehicle for AI and quantum technology companies. In the executive branch, it expands and clarifies the role of the Chief Data Officer and agency data officers, requires recurring state data plans and open-data inventories, and directs agencies to identify high-value data that could be made available for AI, machine learning, and other statistical uses while protecting privacy and avoiding unlawful discrimination.
The bill also directs the Department of Economic and Community Development, with other regulators, to develop a plan for an AI regulatory sandbox that would let companies test innovative products or services under reduced licensing or legal requirements. In addition, it creates a new rule that a person generally cannot avoid civil or administrative liability by arguing that an AI system committed the act or was used in the conduct at issue, while carving out a disclosure-based exception for generative AI when individuals are interacting with a chatbot or similar system, especially in licensed-profession contexts. Finally, it authorizes Connecticut Innovations to establish an Artificial Intelligence and Quantum Technology Investment Fund and invest up to $50 million in companies developing or deploying AI or quantum technology in Connecticut or relocating significant operations to the state.
The bill would amend existing data-governance law in section 4-67p and add several new statutory sections affecting executive branch data practices, economic development policy, consumer protection, professional services, and state investment authority. It would require agencies to formalize data officers, inventories, open-data plans, and compliance with a statewide data plan, while also creating a process for identifying data that can be safely published for AI-related uses. It would further authorize a new AI sandbox planning process and empower Connecticut Innovations to create and capitalize a dedicated investment fund for AI and quantum technology businesses.
The available voting history suggests the bill was received favorably overall, with a 20-1 joint favorable vote. No committee transcript excerpts were provided, so there is no recorded floor or committee debate to indicate broader public arguments, but the strong vote margin suggests substantial support for the bill’s pro-innovation and data-modernization goals. The lone dissent indicates at least some concern, though the record provided does not identify the reason.
The main areas of potential contention are privacy, data-sharing, and liability. The bill requires agencies to identify and potentially publish high-value data for AI and machine-learning uses, but it also directs them to address security, privacy, aggregation, redaction, and discrimination risks, which suggests tension between openness and protection of sensitive information. Another likely point of debate is the liability provision for AI systems, especially the rule that AI involvement is not a defense to civil, administrative, consumer protection, or discrimination claims; businesses may view that as necessary accountability, while others may see it as creating uncertainty. The generative AI disclosure requirements for licensed-profession contexts and the creation of a state-backed investment fund may also raise questions about regulatory burden, consumer protection, and the use of public assets.