top of page
Search

New Connecticut Law Seeks to Regulate AI, Regarding Online Safety, Fair Treatment and Employment

  • Human Rights Research Center
  • 6 hours ago
  • 4 min read

June 4, 2026


HRRC supports the Connecticut administration and governor on the ratification of Senate Bill 5, “An Act Concerning Online Safety,” and calls for other states to implement their own AI legislation. As AI continues to impact human rights through issues of digital safety and protection, HRRC emphasizes the importance of legislative regulation to increase transparency and to limit the dangers of AI systems used in the corporate world.

[Image credit: Markus Winkler via Pexels]
[Image credit: Markus Winkler via Pexels]

On May 29, 2026, Connecticut Governor Ned Lamont ratified a new law on AI known as Senate Bill 5. Following another similar bill, SB 4, which was passed earlier in May to regulate data privacy, the two bills together reshape the state’s online privacy policies. SB 5, however, is more focused on regulating AI in areas such as employment tools, companion chatbots, transparency, data collection, and discrimination.


To protect consumers, providers of subscription-based AI technology must reveal key terms and conditions to consumers before they can use or renew a subscription. Senate Bill 5 also seeks to regulate companion chatbots by imposing protections for users. Companion chatbots can be described as AI companions, which is a kind of artificial intelligence that “provides adaptive, human-like responses to user inputs” and “is able to sustain a relationship across multiple interactions” (Connecticut General Assembly, 2026). Protocols must be in place to address situations in which users express self-harm or violent intentions, and the law restricts companion chatbots from being provided to minors in cases where there is a foreseeable risk that they may encourage harmful behaviors, such as self-harm, suicide, or the unauthorized provision of mental health services.


Moreover, SB 5 establishes obligations for both developers and deployers, or employers, in the use of employment-related AEDTs, or automated employment decision tools, which are technologies that use personal data to generate outputs for employment purposes. Developers must provide deployers with all information required for an AEDT to perform its duties for any AEDT deployed beginning October 1, 2027. In addition, developers must create anonymous reporting channels for deployers’ employees, investigate reports on risks such as algorithmic bias or discrimination, and immediately address those risks to promote AI safety.


Deployers must also disclose to job applicants and employees when they are interacting with an AEDT in cases where its use is not reasonably obvious. Employers are additionally expected to provide workers with details about the types of personal data factored into both the system and its decision-making process.


SB 5 also incorporates anti-discrimination provisions related to AEDTs. Employers can be held liable for discriminatory practices resulting from AEDT systems they implement. Therefore, employers must take into deep consideration the potential for discrimination when entering into contracts with AEDT developers.


With human rights issues on the rise in the digital sector, particularly as AI systems become increasingly common in the corporate world, data privacy, online safety, and fair treatment must be protected more than ever before. While the ratification of this Senate bill is a step forward for Connecticut, nationally, the state joins just a few others, such as Colorado and California, in enacting AI-related protection laws. This suggests the United States still has a long way to go in protecting its citizens in the age of artificial intelligence. However, further efforts to regulate AI may prove challenging as the current Trump administration seeks to block state-level AI regulation.


Glossary

  • Algorithmic bias: refers to prejudicial, discriminatory, unjust, inaccurate, or otherwise disparate performance or outcomes from algorithmic systems based on racial, gender, or other attributes of an individual or a group

  • Artificial intelligence: the capability of computer systems or algorithms to imitate intelligent human behavior 

  • Consumer: one that utilizes economic goods 

  • Corporate: of or relating to a corporation; of, relating to, or being the large corporations of a country or region considered as a unit 

  • Deployers: a person or thing that deploys 

  • Developers: one that develops: such as a person or company that develops computer software 

  • Emergence: the act or an instance of emerging 

  • Enact: to establish by legal and authoritative act specifically : to make into law

  • Governor: one that governs such as: one that exercises authority especially over an area or group: an official elected or appointed to act as ruler, chief executive, or nominal head of a political unit

  • Incorporate: law : to make (the terms of a contemporaneous or earlier document) part of another document by specific reference in that document 

  • Legislation: the action of legislating specifically : the exercise of the power and function of making rules (such as laws) that have the force of authority by virtue of their promulgation by an official organ of a state or other organization

  • Liable: obligated according to law or equity (see equity sense 3) : responsible 

  • Protocols: a system of rules that explain the correct conduct and procedures to be followed in formal situations; a set of conventions governing the treatment and especially the formatting of data in an electronic communications system

  • Provision: the act or process of providing 

  • Ratify: to approve and sanction formally : CONFIRM


References











​Address:

2000 Duke Street, Suite 300, Alexandria, VA 22314, USA

​​Email us:

info@humanrightsresearch.org

Tax exempt 501(c)(3)

EIN: 87-1306523

© 2026 HRRC

bottom of page