United States - New Jersey - Algorithmic Discrimination Guidance (2025)

Overview

The “Guidance on Algorithmic Discrimination and the New Jersey Law Against Discrimination” was issued by the New Jersey Office of the Attorney General and the Division on Civil Rights (DCR) on January 9, 2025. This document clarifies that the existing New Jersey Law Against Discrimination (NJLAD), N.J.S.A. § 10:5-1 et seq., applies to and prohibits “algorithmic discrimination” resulting from the use of artificial intelligence (AI) and other advanced automated decision-making tools. The guidance is part of the DCR's newly launched Civil Rights and Technology Initiative, which aims to address the risks of discrimination and bias-based harassment stemming from advanced technologies. It serves to provide support to the New Jersey public and regulated entities on how the NJLAD applies to the use of AI. While the guidance does not impose new requirements, it reinforces that the NJLAD protects New Jersey residents from algorithmic discrimination and that covered entities can be held liable for such discrimination, even if they did not develop the tool or had no intent to discriminate. The guidance broadly covers the application of NJLAD to automated decision-making tools in contexts including employment, housing, places of public accommodation, credit, and contracting.

The document highlights that automated decision-making tools, while offering potential benefits, also carry risks of algorithmic discrimination if not designed and deployed responsibly. It emphasizes that the NJLAD's broad remedial purpose of eliminating discrimination in New Jersey means it draws no distinctions based on the mechanism of discrimination, whether human-driven or facilitated by automated tools. The guidance specifically addresses how bias can be introduced or amplified during the design, training, and deployment phases of AI tools, leading to disparate treatment, disparate impact, or the failure to provide reasonable accommodations.

Definitions

The guidance defines key terms to clarify its scope and application:

• Algorithmic Discrimination: This is defined as discrimination that results from the use of automated decision-making tools. It encompasses discriminatory outcomes that may arise from the design, training, or deployment of such tools.
• Automated Decision-Making Tools: These generally refer to any technological tool, including, but not limited to, a software tool, system, or process that is used to automate all or part of the human decision-making process. Examples provided include generative AI, machine-learning models, traditional statistical tools, and decision trees.
• Covered Entity: An entity subject to the requirements of the New Jersey Law Against Discrimination (LAD). This includes employers, housing providers, places of public accommodation, credit providers, and contractors.
• Disparate Treatment Discrimination: Occurs if a policy or practice is intentionally discriminatory, or if it is discriminatory on its face, even without intent. In the context of automated tools, it happens when a covered entity designs or uses such tools to intentionally treat members of a protected class differently.
• Disparate Impact Discrimination: Occurs when automated decision-making tools recommend or contribute to decisions that disproportionately affect members of a protected characteristic, regardless of intent, unless the use of the tool serves a substantial, legitimate, nondiscriminatory interest and there is no less-discriminatory alternative.

Governance and Institutional Framework

The primary governmental bodies responsible for this guidance and its enforcement are the New Jersey Office of the Attorney General (NJ OAG) and the New Jersey Division on Civil Rights (DCR). The DCR is an agency within the NJ OAG. The guidance itself was issued by Attorney General Matthew J. Platkin and the DCR Director.

As part of this initiative, the DCR has launched a new Civil Rights and Technology Initiative. A key component of this initiative is the creation of a Civil Rights Innovation Lab. This lab aims to leverage technology responsibly to advance the DCR's mission to prevent, address, and remedy discrimination. The Innovation Lab is intended to work with experts, technologists, and stakeholders to identify and develop technology that enhances the DCR's enforcement, outreach, and public education efforts. It will also develop protocols to facilitate the responsible deployment of such technology. The DCR's Civil Rights and Technology Initiative is informed by the recommendations from Governor Phil Murphy's Artificial Intelligence Task Force, which emphasized the need for public education on bias and discrimination related to AI deployment.

Key Focus Areas

• Prohibition of Algorithmic Discrimination: The core of the guidance is the clarification that the New Jersey Law Against Discrimination (NJLAD) prohibits algorithmic discrimination. This means that discrimination resulting from the use of automated decision-making tools is unlawful, just as discrimination by human actors is. The NJLAD prohibits discrimination based on actual or perceived race, religion, color, national origin, sexual orientation, pregnancy, breastfeeding, sex, gender identity, gender expression, disability, and other protected characteristics.
• Scope of Application: The guidance applies to algorithmic discrimination stemming from the use of AI and other automated decision-making tools by entities covered under the LAD. These include employers, housing providers, places of public accommodation, credit providers, and contractors.
• Sources of Algorithmic Bias: The guidance identifies three primary stages where bias can be introduced into automated decision-making tools, leading to discriminatory outcomes:
  • Design: Bias can stem from the tool's model, algorithm, or the inputs chosen by the developer, either intentionally or inadvertently.
  • Training: If the datasets used to train AI tools contain historical biases or reflect systemic inequities, the tool may perpetuate or amplify those biases.
  • Deployment: Discriminatory results can occur even with unbiased tools if they are inconsistently used, misused, or applied in contexts for which they were not designed.
• Types of Discrimination Covered: The NJLAD's protections extend to algorithmic discrimination that results in disparate treatment, disparate impact, or the failure to provide reasonable accommodations based on protected characteristics.
• Entity Responsibility: The guidance clarifies that covered entities are responsible for the AI technology they utilize and cannot delegate their compliance responsibilities to third parties, such as tool developers or vendors. An entity can be held liable even if it has no intent to discriminate or does not fully understand the inner workings of a third-party tool.
• Public Education and Outreach: The Civil Rights and Technology Initiative, including the Civil Rights Innovation Lab, emphasizes public education on bias and discrimination related to AI deployment. The DCR plans to make training available to the public on bias and discrimination from AI tools.

Implementation Framework

The guidance itself does not establish new legal requirements but rather clarifies how existing provisions of the New Jersey Law Against Discrimination (NJLAD) apply to the use of automated decision-making tools and artificial intelligence. Therefore, entities covered by the NJLAD are expected to comply with these interpretations immediately upon the guidance's issuance on January 9, 2025. The DCR's goal is to provide clarity and support for compliance with the LAD.

Implementation involves covered entities reviewing their use of AI and automated decision-making tools to ensure they align with the NJLAD's prohibitions against discrimination. This includes assessing their tools for potential biases in design, training, and deployment. The guidance suggests that employers should understand the AI tools they use, train employees on their correct usage, and audit tools for biases. Although the guidance does not impose a mandatory audit requirement akin to some other state laws (e.g., New York City's law for automated employment decision tools), it strongly implies the necessity of such practices for compliance and risk mitigation. The DCR also announced plans to host virtual programs and provide free public training on AI bias and discrimination later in 2025 to aid in understanding and compliance.

Monitoring and Evaluation

The New Jersey Division on Civil Rights (DCR), under the Office of the Attorney General, is responsible for monitoring compliance with the New Jersey Law Against Discrimination (NJLAD), which now explicitly includes algorithmic discrimination in its interpretive scope. The DCR’s Civil Rights and Technology Initiative and the Civil Rights Innovation Lab are key institutional frameworks for this monitoring and evaluation. The Innovation Lab is specifically tasked with identifying and developing technology to enhance DCR's enforcement, outreach, and public education work, and to develop protocols for the responsible deployment of technology.

The guidance itself does not lay out specific reporting requirements for entities using AI. However, it implicitly suggests that entities should proactively evaluate their AI tools. Recommendations for best practices include ensuring a thorough understanding of AI tools, training employees on their correct use, and regularly auditing for biases and disproportionate impacts. The DCR's enforcement will likely involve investigating complaints of algorithmic discrimination under the existing framework of the NJLAD. The focus remains on the outcome of the automated decision-making process; if it results in unlawful discrimination, the entity is liable.

Penalties, Liability, and Appeals

The guidance clarifies that liability for algorithmic discrimination falls under the existing New Jersey Law Against Discrimination (NJLAD). Therefore, the penalties, liability standards, and appeals processes established by the NJLAD apply directly to cases of algorithmic discrimination. The NJLAD is a comprehensive civil rights law that prohibits discrimination in various contexts, including employment, housing, public accommodations, credit, and contracting.

Key aspects of liability under the guidance include:

• Strict Liability: An entity can violate the NJLAD even if it has no intent to discriminate. The focus is on the discriminatory outcome, not the intent behind the use of the automated tool.
• Third-Party Liability: Covered entities may be held liable for algorithmic discrimination even if they use an automated decision-making tool developed by a third party. Employers cannot shift liability to their AI vendors or external developers. This underscores the responsibility of the entity deploying the AI to ensure its compliance.
• Types of Discrimination: Liability can arise from disparate treatment discrimination (intentional or facially discriminatory design/use), disparate impact discrimination (disproportionate adverse effect without a legitimate, nondiscriminatory interest and no less-discriminatory alternative), or the failure to provide reasonable accommodations due to algorithmic decisions.

The specific penalties under the NJLAD can include monetary damages, injunctive relief (ordering the cessation of discriminatory practices), and other remedies deemed appropriate by the Division on Civil Rights or the courts. The DCR has the authority to investigate complaints, make findings, and issue orders to remedy discrimination. Parties aggrieved by DCR decisions may have avenues for appeal through the administrative and judicial systems, as provided by the NJLAD and New Jersey's administrative procedure laws. The guidance reinforces that claims of algorithmic discrimination will be assessed consistent with other claims of discrimination under the LAD.

Relationship to Other Instruments

The “Guidance on Algorithmic Discrimination and the New Jersey Law Against Discrimination” is intrinsically linked to and serves as an interpretive document for the existing New Jersey Law Against Discrimination (NJLAD), N.J.S.A. § 10:5-1 et seq.. The guidance explicitly states that it does not impose any new or additional requirements beyond what is already included in the NJLAD, nor does it establish new rights or obligations; rather, it clarifies how the LAD applies to algorithmic discrimination. The DCR will enforce the NJLAD, not the guidance document itself as a substitute for the law.

This guidance also operates within the broader context of other New Jersey legislative and governmental actions related to AI and data protection. For instance, the DCR's initiative is informed by the work of Governor Murphy's Artificial Intelligence Task Force, which issued a report on AI. Furthermore, New Jersey enacted a law against deceptive AI deepfakes (Bill A3540) on April 2, 2025, establishing civil and criminal penalties. There is also the New Jersey Data Privacy Act (NJDPA), signed into law on January 16, 2024, and effective January 15, 2025, which regulates the processing of personal data and includes provisions related to profiling and automated decision-making, requiring data protection assessments for high-risk processing. Proposed regulations under the NJDPA further narrow exemptions for internal research, particularly concerning AI model training, and require affirmative consent for certain data uses. Other legislative efforts in New Jersey include bills concerning AI use by state agencies (S1438), promoting AI industry development (S3432/A4558), and regulating AI in the news media (A5164). This guidance complements these efforts by addressing the civil rights implications of AI under existing anti-discrimination law.

International Alignment

While the New Jersey Algorithmic Discrimination Guidance is a state-level document, its underlying principles align with broader national and international conversations and emerging regulatory trends regarding AI and anti-discrimination. The guidance itself cites the U.S. Equal Employment Opportunity Commission's (EEOC) “Select Issues: Assessing Adverse Impact in Software, Algorithms, and Artificial Intelligence Used in Employment Selection Procedures Under Title VII of the Civil Rights Act of 1964”. This indicates an awareness and alignment with federal interpretations of anti-discrimination laws in the context of AI.

Globally, there is a growing consensus on the need to address algorithmic bias and discrimination to protect fundamental rights. The European Union's General Data Protection Regulation (GDPR), for example, includes provisions related to automated individual decision-making, and many emerging AI regulations, such as the EU AI Act, focus heavily on risk management, transparency, and fundamental rights in AI systems. While not directly adopting international standards, New Jersey's guidance reflects a similar concern for protecting individuals from discriminatory outcomes caused by AI, emphasizing principles like fairness, accountability, and the non-discriminatory application of technology. The guidance's focus on disparate impact and disparate treatment echoes legal doctrines found in many jurisdictions that aim to prevent both intentional and unintentional discrimination, regardless of the mechanism employed. This positions New Jersey alongside other jurisdictions worldwide that are actively working to ensure that technological advancements do not erode civil rights protections.

Implementation Timeline

Compliance Checklist

Sources and References