Researcher: DC Page 1 4/9/25
OLR Bill Analysis
SB 1249
AN ACT ADDRESSING INNOVATIONS IN ARTIFICIAL
INTELLIGENCE.
SUMMARY
This bill establishes various artificial intelligence (AI)-related
reviews, programs, and funds. It also specifies that it is generally not a
defense to any civil or administrative claim or action that an AI system
committed or was used in furthering the act or omission the claim or
action is based on.
Specifically, the bill requires the following:
1. the chief data officer to review the inventory of high value data
and identify and publish any that could be useful for AI systems,
machine learning, and other statistical means of data analysis;
2. the economic and community development (DECD)
commissioner to develop a plan to establish an AI regulatory
sandbox program (i.e. one that allows for temporary testing of
products or services with fewer legal requirements); and
3. Connecticut Innovations, Inc. (CI) to establish the “Artificial
Intelligence and Quantum Technology Investment Fund” to
invest in Connecticut companies developing, manufacturing,
using, or deploying AI systems or quantum technology.
EFFECTIVE DATE: July 1, 2025, except the defense prohibition
provision is effective October 1, 2025.
§ 1 ─ HIGH VALUE DATA ANA LYSIS FOR AI USE
By January 1, 2026, the bill requires the chief data officer, in
consultation with agency data officers, to review the inventory of high
value data that executive branch agencies are required to have or collect 2025SB-01249-R000606-BA.DOCX
Researcher: DC Page 2 4/9/25
under existing law. Under existing law, “high value data” is any data
that the department head determines:
1. can increase an agency’s accountability and responsiveness,
improve public knowledge of an agency and its operations,
further its core mission, or create economic opportunity;
2. is critical to the agency’s operation or used to satisfy any
legislative or other reporting requirements; or
3. is frequently requested by the public or responds to a need and
demand identified through public consultation.
Under the bill, the chief data officer must identify and publish any
data that could be useful for AI systems, machine learning, and other
statistical means of data analysis. He must do so to create economic
opportunity and support state economic development goals, through
private businesses, nonprofit organizations, and other entities that will
use the data consistent with applicable laws and regulations.
In reviewing the data, the chief data officer and agency data officers
must:
1. identify appropriate data to make available for use by AI
systems, machine learning, and other statistical means of data
analysis;
2. develop policies and procedures for data quality and governance
to ensure data are appropriate for the intended purpose and do
not lead to any unlawful discrimination against any person or
group of people or disparate impact on any individual or group
based on any actual or perceived differentiating characteristic
outlined in state law (e.g., age, race, disability);
3. determine any needed aggregation, redaction of individually
identifiable information, or use of other techniques needed to
ensure and preserve privacy and to satisfy all applicable state or
federal laws and regulations for publicly disclosing data; and 2025SB-01249-R000606-BA.DOCX
Researcher: DC Page 3 4/9/25
4. determine the procedures for agencies to publish their data on
the online repository that the Office of Policy and Management
operates and maintains under existing law.
§ 2 ─ AI REGULATORY SANDBOX PROGRAM
The bill requires the DECD commissioner, in consultation with the
banking, health strategy, public health, and insurance commissioners,
to develop a plan to establish an AI regulatory sandbox program. The
program must allow an applicant to temporarily test an innovative
product or service on a limited basis under reduced licensure,
regulatory, or other legal requirements.
The plan must be developed to make Connecticut’s business
environment competitive, relative to other places, for developing and
deploying AI technologies.
Not later than January 1, 2026, the DECD commissioner must submit
recommendations for any required legislative proposals for
implementing the plan to the governor and the Commerce, Banking,
Insurance and Real Estate, and Public Health committees.
§ 3 ─ DEFENSE TO CIVIL OR ADMINISTRATIVE ACTION
The bill specifies that it is generally not a defense to any civil or
administrative claim or action that an AI system committed or was used
in furthering the act or omission the claim or action is based on.
These claims or actions include tort or contract claims; Connecticut
Unfair Trade Practices Act (CUTPA) claims or actions; unauthorized
practice of any licensed, certified, or registered profession, trade,
business, or occupation claims or actions; or discrimination claims or
actions under the human rights and opportunity laws.
For this provision, an “AI system” means any machine-based system
that, for any explicit or implicit objective, infers from the inputs the
system receives how to generate outputs, including content, decisions,
predictions, or recommendations, that can influence physical or virtual
environments. 2025SB-01249-R000606-BA.DOCX
Researcher: DC Page 4 4/9/25
Exception
However, it can be a defense to a claim or action when a person (i.e.
individual or entity) giving someone access to a generative AI system,
or prompting or causing an individual to interact with these systems,
gives certain disclosures to the individual.
The person must clearly and conspicuously disclose that the
individual is interacting with a generative AI system and not a human
being (1) when the individual requests the disclosure, or (2) orally at the
beginning of an oral interaction or, for written interactions, by an
electronic message delivered to the individual before the interaction
begins.
If the system is provided by a credentialed person delivering services
within the scope of the applicable licensed, certified, or registered
profession, trade, business, or occupation, they must also disclose that
the interaction between the individual and the system is not intended to
give any advice that only a person acting within their credential’s scope
may give.
Under the bill, a “generative AI system” means any AI system that
(1) is trained on data; (2) interacts with any individual through text,
audio, or visual communication; and (3) generates non-scripted outputs
similar to human outputs with limited or no human oversight.
§ 4 ─ ARTIFICIAL INTELLIGENCE AND QUANTUM TECHNOLO GY
INVESTMENT FUND
The bill requires CI to establish a fund to be known as the “Artificial
Intelligence and Quantum Technology Investment Fund” for the
purpose of investing in Connecticut companies, or companies relocating
significant operations to Connecticut, which develop, manufacture, use,
or deploy AI systems or quantum technology.
For the purposes of this fund, an “AI system” has the same meaning
as described above. And “quantum technology” means the use of the
laws of quantum physics for the (1) storage, transmission, manipulation,
computing, or measurement of information, or (2) development or
manufacturing of machinery, hardware, materials, or pharmaceuticals. 2025SB-01249-R000606-BA.DOCX
Researcher: DC Page 5 4/9/25
Under the bill, CI or its subsidiary must serve as the fund’s general
partner or managing member and determine the fund’s organizational
and operating structure. CI may (1) capitalize the fund by drawing from
existing assets and (2) invest up to $50 million in the fund. The fund may
make investments or co-investments in Connecticut companies and
those relocating significant operations to the state. The investments or
co-investments may include seed, start-up, early, first-stage, or
expansion financing.
BACKGROUND
Related Bills
sSB 2, favorably reported by the General Law Committee, establishes
a framework for regulating artificial intelligence and includes other AI-
related provisions, including establishing a regulatory sandbox (§ 12).
sSB 10, § 5, favorably reported by the Insurance and Real Estate
Committee, prohibits health carriers from using artificial intelligence or
algorithms in place of a clinical peer to evaluate the clinical
appropriateness of an adverse determination.
sSB 1484, favorably reported by the Labor and Public Employees
Committee, imposes limits on an employer’s use of high-risk AI systems
to make consequential decisions by, among other things, requiring
employers to have an impact assessment before deploying it and giving
employees certain information about the systems and how they are
used.
HB 6846, favorably reported by the Government Administration and
Elections Committee, generally makes it a crime for a person to, 90 days
before an election or primary, (1) distribute certain communication with
deceptive synthetic media or (2) enter into an agreement to distribute it.
COMMITTEE ACTION
General Law Committee
Joint Favorable
Yea 20 Nay 1 (03/21/2025)