Provides disparate impact based on automated decision system as cause of action for certain consumers.
Senate Bill 4279 would regulate the use of automated decision systems, including artificial intelligence, machine learning, predictive analytics, and related data tools, when they are used in housing and credit decisions. The bill applies to covered entities such as lenders, housing providers, screening companies, creditors, servicers, insurers, agents, and vendors involved in decisions about mortgage credit, rental housing, rental pricing, and related financial services. It declares that using such systems in a way that produces a disparate impact on a protected class would violate the New Jersey Law Against Discrimination.
The bill creates a private right of action in Superior Court for individuals who believe they were harmed by an automated decision system. To defend against liability, a covered entity would have to show that the challenged use serves a substantial, legitimate, nondiscriminatory purpose and that no less discriminatory alternative is available. The bill also requires covered entities to conduct algorithmic impact assessments at least every two years, keep records for inspection by the Division on Civil Rights or the Attorney General, provide notice when an automated system is used, and allow meaningful human review when requested. It also makes covered entities liable for violations by agents, third-party vendors, model developers, or external data sources involved in the decision-making process.
The bill would supplement Title 10 and expand the New Jersey Law Against Discrimination framework to expressly cover disparate-impact claims arising from automated decision systems in housing and credit contexts. It would impose new compliance duties on entities using these systems, including periodic impact assessments, recordkeeping, consumer notice, and human-review rights, while authorizing enforcement by the Division on Civil Rights and the Attorney General through fines and existing LAD remedies. It would also expose covered entities to liability for the conduct of downstream vendors and developers involved in the automated system.
The bill’s stated purpose and structure reflect a strong consumer-protection and civil-rights orientation, with the sponsor framing the measure as necessary to preserve fair access to housing and credit in light of growing use of AI and reduced federal disparate-impact enforcement. No committee transcript or vote record is available in the provided materials, so there is no documented legislative debate or recorded vote sentiment to assess beyond the bill text and sponsor statement.
The main points of contention are likely to be the bill’s creation of a new disparate-impact cause of action for automated systems, the compliance burden of mandatory algorithmic impact assessments, and the scope of liability extending to agents, vendors, model developers, and external data sources. Covered entities may also object to the potential exposure to litigation and fines, while supporters are likely to emphasize transparency, accountability, and the need to prevent discriminatory outcomes in housing and lending. The bill attempts to address industry concerns by preserving proprietary source code and trade secrets from disclosure and by providing an affirmative defense where no less discriminatory alternative exists.