Employers need to be smarter than ever about how they use artificial intelligence (AI) in the workplace. Laws attempting to regulate the use of AI in the workplace have seemingly kept pace with advancements in the technology itself. Originally intended to streamline employment processes, AI may have unintended consequences that need to be mitigated. There is nothing artificial about the risk employers may face if they are not transparent and careful about how they use AI to make employment decisions. Employers must review their vendor agreements and AI protocols carefully, and conduct regular audits of AI’s processes, to minimize liability and avoid misuse or discriminatory outcomes, while maximizing the benefit of greater efficiencies in the workplace which properly used AI tools may deliver.

Case Law Catch-Up

Over the summer, a federal court left open the possibility that an employer could be directly liable under Title VII, the Americans with Disabilities Act (ADA), and the Age Discrimination in Employment Act (ADEA) for the biased AI practices of vendor(s) it engages to screen applicants. That case involved a putative class of job applicants claiming that the AI vendor’s applicant screening tools discriminated on the basis of race, age, and disability in violation of federal and state antidiscrimination laws. The court rejected the AI vendor’s argument that it was only providing a tool that implemented the employers’ criteria, and instead found that the vendor, through its human capital management platform, “participat[ed] in the decision-making process by recommending some candidates to move forward and rejecting others.” The court allowed the plaintiffs’ claims to proceed against the AI vendor under the theory that the vendor was either an agent of its client-employers or was delegated functions traditionally exercised by employers. With the claims given the green light to proceed, it is possible that discovery will test the parameters of this new employer-agent theory.

Legislative Catch-Up

On the legislative side, state and local governments have continued to make waves in AI regulation.

  • For starters, New York City’s trailblazing law regulating the use of automated employment decision tools (AEDTs) and requiring employers to ensure those tools undergo bias audits went into effect on July 5, 2023.
  • More recently, New York State introduced two bills, S7623 and A9315, which would both amend the NY Labor Law to make it unlawful to use any “electronic monitoring tool” (much more broadly defined than AEDTs) unless certain conditions are met.
  • Colorado also recently signed SB 205 into law, which will go into effect in February 2026. SB 205 has been lauded as the most comprehensive law to date governing employers’ use of AEDTs. While the Colorado law imposes similar requirements to New York City’s version, it will additionally require employers to take action in response to bias audits, and require tech developers to take “reasonable care” to protect residents from discrimination (i.e., by sharing information about the AI processes with their employer-clients). The Colorado law also adds a requirement that employers provide a notice and appeals process for every single applicant negatively affected by the AI-assisted employment decision.
  • Illinois, which already requires employers that use AI to evaluate video interviews to notify applicants and obtain their consent for the use of AI, also just signed H.B. 3773 to amend the Illinois Human Rights Act and make it unlawful for employers to use discriminatory AI technology in recruitment, hiring, promotion, discipline, termination and other terms, privileges, or conditions of employment. A different Illinois bill in the works, H.B. 5116, would require employers to conduct bias audits of the AEDTs they use.

While this is just a sampling of recent legislative developments, it exemplifies the trend of AI regulation presenting itself in previously unchartered territory.

Wading Through Developing Waters

As these recent updates reflect, employers who use any level of non-human decision-making in any aspect of the pre-employment and employment recruiting, screening, interviewing, and rating process should proceed with caution and awareness of the adapting legal landscape.

In addition to ensuring that New York (and eventually Colorado) employers notify individuals about the use of AI and conduct bias audits at the required frequencies, employers should have comprehensive knowledge about how the AI tools they use work, and what role the AI plays in their employment decisions. These safeguarding measures will help employers defend those decisions down the stream, if necessary. Understanding the design, programming, and training that drive an AI tool’s algorithms can help mitigate the risk of bias reflected in audits — and as a result, litigation based on bias. For example, an AI tool’s training data may mislabel data or misrepresent certain groups, or contain errors in its code, all of which can contribute to biased outputs. Catching those errors or inherent biases from the outset may be key to avoiding a class action lawsuit.

Paying close attention to the results of bias audits will also be key for employers moving forward. If an audit reveals disparate impacts caused by the AI, an employer should work with its vendor on ways to mitigate the bias, such as modifying the training data or algorithm. Employers may also consider using another tool, or ceasing use of AI tools altogether.

Employers should also be mindful of the need to provide reasonable accommodations to applicants and employees who request to opt out of the AI review process, and be prepared for human review or other options.

Swimming Ahead of the AI Current

Now is also a great time for employers to reevaluate the vendors they use and renegotiate any contracts that do not shift any (or enough) liability to the vendor for any issues it causes or contributes to vis-à-vis the use of an AI tool. Employers should strive for the inclusion of an indemnification provision in any AI vendor contract to the extent possible. Employers should also create and make sure that any AI policies are up-to-date, and that their HR, Talent, IT, and any other teams are well versed on the AI tool(s) being used and receive all necessary discrimination training. When used correctly and responsibly, AI tools can be a valuable resource and life jacket for employers in a variety of areas.

For guidance on navigating AI issues in the workplace, consult your Akerman labor and employment attorney.