Building Responsibly: What Employers Need to Know About AI Compliance in the Workplace

This article was updated on October 21, 2026 to reflect the latest guidance.

Artificial intelligence (AI) is fundamentally reshaping virtually every industry, and the workplace is no exception.^1,2 Through conversations with our HR Consulting teams, it has become increasingly clear that what once seemed like a distant concept is now a daily reality in talent acquisition, performance reviews, and workforce planning. But astechnologies continue to advance, so too does the responsibility to deploy them thoughtfully. In response, federal, state, and local lawmakers are pursuing measures to ensure AI use in the workplace aligns with existing employee rights and protections.

The Federal Landscape: The Push for a National Framework 

At the federal level, the regulatory landscape for AI has shifted considerably. Rather than building on prior federal guidance, much of which has been revoked or deprioritized, the Trump administration has moved decisively toward a deregulatory approach that prioritizes AI innovation over oversight.

That shift began in January 2025, when President Trump signed Executive Order 14179, signaling a move away from the more restrictive AI guidance of prior years and emphasizing instead United States’ leadership in AI development. That deregulatory posture was reinforced in December 2025 when President Trump signed Executive Order 14365, directing federal agencies to use lawsuits and funding cuts to challenge state-level AI regulations deemed to be inconsistent with federal policy. 

Most recently, in March 2026, the Trump administration released “A National Policy Framework for Artificial Intelligence,” providing a non-binding set of legislative recommendations that urge Congress to adopt a unified federal approach to AI regulation. The framework flows from EO 14365 and addresses six priority areas, including safeguards for small businesses, preservation of intellectual property rights, and expansion of AI-related workforce opportunities. It is important to note, however, that Congressional action would be required before any of the framework’s recommendations become enforceable regulatory obligations; an outcome that remains uncertain. In the meantime, the divergence between federal deregulation and state-level compliance requirements means employers cannot rely on the federal framework alone.

States Take the Lead  

While the federal government works through its own transition, states have not waited. Across the country, legislators are advancing their own frameworks for responsible AI use in the workplace, each taking different approaches to balancing innovation with employee protections. 

California has emerged as a national leader in AI regulation. The California Civil Rights Department approved final regulations clarifying how existing anti-discrimination laws apply to automated decision systems in employment, which took effect October 1, 2025. The rules confirm that it is unlawful to use AI and automated-decision making tools to make employment-related decisions that discriminate against applicants or employees and reaffirm that established criminal history and medical inquiry restrictions continue to apply even when decisions are made by or with the assistance of AI technology. Covered systems include tools such as those that screen resumes, rank applicants, analyze facial expressions or voice, and target job ads. Employers remain responsible for compliance even when using third-party vendors or technology platforms. 

California has also made clear that accountability for AI-driven outputs rests with the humans and organizations that develop, modify, or deploy the technology, meaning employers cannot disclaim responsibility by arguing that “AI acted on its own.” Reinforcing this principle, California’s Transparency in Frontier Artificial Intelligence Act (TFAIA) also imposes new disclosure and internal reporting standards for large-scale AI developers and provides whistleblower protections for employees who report safety risks or violations to authorities.  

In May 2026, Governor Newsom signed Executive Order N-6-26, directing state agencies to study AI's effect on California's labor market, launch a dashboard tracking AI's employment impact using unemployment insurance data, and recommend updates to the state's WARN Act. The order imposes no obligations on employers, but it signals how the state may look to balance AI innovation with worker protections.

Looking ahead, California’s Privacy Protection Agency (CPPA) has also finalized regulations under the California Consumer Privacy Act addressing the use of automated decision-making technology (ADMT) in employment decisions. Set to go into effect on January 1, 2027, the CPPA’s regulations require covered businesses to provide pre-use notices with opt-out options and allow employees access to ADMT logic. The regulations also impose specific risk assessment and reporting obligations for employers who use ADMTs for “significant” employment decisions. 

The California Legislature also passed several AI measures set to take effect in 2027 and beyond. The most employer-facing are the No Robo Bosses Act (SB 947), which bars employers from relying solely on an automated decision system for discipline or termination and requires human corroboration where such a system is relied on primarily, and a technological displacement notice requirement (SB 951) that expands what employers must include in CalWARN notices when layoffs are caused in substantial part by automation. The remaining measures address developers, auditors, and attorneys: a registration system for AI auditors (AB 1405), independent verification organizations (SB 813), and new limits on attorneys' use of generative AI (SB 574).

Colorado was the first state to pass a comprehensive AI anti-discrimination law, in 2024. After a good deal of back and forth (a court challenge and a legislative rewrite), the revised law is set to take effect on January 1, 2027. The law applies to employers that use automated decision-making technology (“ADMT”) to materially influence consequential employment decisions, including hiring, promotion, compensation, and termination. It requires pre-use notice, a post-adverse-action process with an opportunity for human review, and a three-year recordkeeping obligation. While the revised law removes, among other requirements, its original impact assessment, risk management, and annual review obligations, it leans on existing state and federal anti-discrimination law to hold employers accountable for discriminatory outcomes. Liability cannot be shifted to a developer or tool vendor, meaning employers remain liable for discriminatory outcomes that result from the technology they use.

Colorado’s attorney general proposed implementing rules in August 2026, the most consequential being employers’ obligation to explain its use of ADMTs for covered decisions. As proposed, when an AI tool contributes to an adverse outcome, an employer would have to explain in clear, specific terms, the tool's specific purpose, the role it played in the decision, and the role of any human reviewer. This means that employers must understand how their AI tools actually work. An employer that cannot explain the tool's role in a decision will not be able to comply.

Connecticut is the newest addition to the list. SB 5, which took effect on October 1, 2026, applies to employers that use automated employment decision tools (“AEDT”) to support decisions about hiring, promotion, discipline, and termination. It requires employers to disclose when AI interacts with an applicant or employee in real time and to provide written notice before an AEDT-supported employment decision is made. Although the law does not impose an independent bias audit requirement, it does indicate that evidence of bias testing may be relevant during an enforcement action. A pilot program for independent verification of covered tools is also set to begin July 1, 2027.

The starting point is to understand where and how AI is currently being used within your organization. This includes developing an understanding of where the tool draws data from, how it is trained to evaluate that data, and how it ultimately generates output. 

Illinois has adopted a multi-pronged approach. The state’s AI Video Interview Act, in effect since 2020, requires employers to obtain consent and report demographic data when using AI to analyze video interviews. A further restricts the use of digital replicas in employment contracts. Most recently, HB 3773, which took effect on January 1, 2026, amends the Illinois Human Rights Act in two ways: it requires employers to notify individuals when AI is used to support employment decisions, from recruitment through discipline and discharge, and it confirms that employers are liable when that use results in unlawful discrimination. Separately, SB 315, effective January 1, 2027, imposes transparency, audit, and incident-reporting duties on large-scale frontier AI developers, with third-party audits beginning in 2028.

New Jersey launched its Civil Rights and Technology Initiative in 2025, through which the state’s Attorney General issued guidance clarifying that algorithmic bias violating the New Jersey Law Against Discrimination is prohibited. Employers using AI tools must proactively test for bias and ensure that reasonable accommodations are not overlooked, regardless of whether discriminatory outcomes were intentional.  

Texas enacted the Texas Responsible Artificial Intelligence Governance Act (TRAIGA) which establishes a regulatory framework for AI systems in the state. While TRAIGA does not impose direct obligations on private employers, it does prohibit the development or deployment of AI systems that intentionally discriminate against protected classes in violation of federal or state law.  

New York’s Responsible AI Safety and Education Act (RAISE Act), set to take effect January 1, 2027, imposes safety and transparency requirements on developers of large-scale “frontier models” that meet state-defined compute and revenue thresholds. Covered developers must publish safety frameworks and transparency reports, file quarterly risk summaries, and report critical safety incidents to a new oversight office within the state's Department of Financial Services. Separately, New York’s FutureWorks Commission, launched in March 2026, is examining AI's impact on the state's workforce and will report findings by the end of the year. Its recommendations will shape the next round of AI regulation in New York, which is expected to reach employers directly. While the RAISE Act shares structural similarities to California’s TFAIA, it establishes its own distinct standards, underscoring the growing complexity of the state-by-state AI compliance landscape.

Local Governments Add Another Layer  

At the local level, cities are also adding their own requirements to the growing framework of responsible AI use in employment. New York City’s Local Law 144, which took effect in 2023, requires employers to conduct independent bias audits of automated employment decision tools and provide transparent notice to applicants and employees about their use. Additionally, Portland, Oregon, has banned the use of facial recognition technology in “any place of public accommodation.” 

What Can Employers Do Now?  

Whatever form federal policy ultimately takes, the common thread running through nearly every state regulation issued to date is clear: the use of AI does not excuse compliance with existing anti-discrimination, employee privacy, or any other applicable employee protection law. 

The starting point is to understand where and how AI is currently being used within your organization. This includes developing an understanding of where the tool draws data from, how it is trained to evaluate that data, and how it ultimately generates output. Armed with that understanding, employers can begin to assess whether those tools are operating in a manner that is consistent with regulatory requirements.  

To build a responsible foundation, employers should consider auditing their AI tools for bias, implementing transparency practices regarding AI-driven decisions that affect applicants and employees, and training HR and management personnel on responsible AI use. As AI regulation continues to evolve, staying informed is one of the most valuable steps an employer can take. Employers are encouraged to monitor developments within their jurisdictions to ensure their AI practices remain aligned with applicable requirements. 

---

References

[1] https://reports.weforum.org/docs/WEF_Future_of_Jobs_Report_2025.pdf

[2] https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai