Employment Practices Liability

AI in Hiring and Employee Monitoring: Algorithmic Discrimination and EPLI Questions

EPLI policies typically respond to discrimination claims regardless of whether a human or an algorithm made the decision, because the policy covers the allegation of discrimination, not the method behind it. The harder questions are whether the policy also covers privacy claims from employee monitoring, whether your vendor shares any of the liability, and whether you can defend a decision when you cannot explain how the tool reached it.

Employers are adopting artificial intelligence throughout the employment lifecycle. Resume screeners rank applicants before a recruiter sees them, video-interview platforms score candidates on word choice and expression, and productivity software tracks keystrokes, location and messaging. The goal is not to avoid these tools but to use them in a way that holds up if challenged, and to confirm the insurance program is built for the claims they can generate.

How algorithmic bias becomes a claim

Most AI hiring tools learn from historical data. If a company's past hires skewed toward a particular group, a model trained on those hires may learn to prefer candidates who resemble them. The tool does not need to consider race, sex or age directly; it can pick up proxies such as graduation year, zip code, employment gaps or the name of a women's college.

This is where disparate-impact theory comes in. A policy or practice that appears neutral can still be unlawful if it disproportionately screens out a protected group and the employer cannot show it is job-related and consistent with business necessity. The employer does not need to intend discrimination, and "the software did it" is not a defense; the employer that chose and deployed the tool is generally the responsible party.

Video-interview scoring raises additional concerns. Tools that analyze speech patterns or facial movement may disadvantage candidates with disabilities, accents or cultural differences in expression, which can support disability or national-origin claims on top of the disparate-impact theory.

Some states and cities have begun regulating automated employment decision tools. Common themes include notifying candidates that a tool is in use, offering an alternative process on request, and conducting periodic bias audits. Requirements vary and continue to change, so confirm what applies to you with employment counsel or your state labor agency. Even where no rule applies, documented notice and periodic testing are becoming the expected standard.

Employee monitoring and privacy claims

Monitoring software introduces a different category of exposure. Tracking keystrokes, screenshots, location or communications can trigger claims under state privacy laws and electronic communications statutes, and in some cases disability or labor law if the data is used to discipline employees for protected activity or medical absences.

Many states require notice to employees before electronic monitoring begins, and some require written acknowledgment. Workers may also allege that monitoring data was used in a discriminatory way, for example if productivity scores that penalize caregivers or employees with disabilities feed into performance reviews or layoff decisions.

Whether EPLI responds depends heavily on policy wording. Discrimination and retaliation allegations tied to monitoring data are typically covered. A standalone invasion-of-privacy claim may be covered if the policy lists it as a wrongful employment act, and many do, but some policies exclude violations of specific privacy or biometric statutes, and statutory penalties are often excluded.

Vendor contracts and third-party E&O

Your contract with the AI vendor is your first line of defense, and it is often weaker than you assume. Many agreements disclaim responsibility for how the customer uses the tool and cap the vendor's liability at a year of fees. Ask for a representation that the tool has been tested for adverse impact, a commitment to share audit results, and an indemnity for claims arising from defects in the model itself.

Also ask the vendor for a certificate showing its own technology errors and omissions coverage, and consider requiring it as a contract term. Vendor E&O may provide a source of recovery if the model turns out to be flawed, but it generally will not defend you in an employment lawsuit. Your EPLI policy is what responds when an applicant names your company, so the two coverages serve different purposes.

Documentation and human review

The most effective risk control is also the simplest: keep a human in the loop and write down what the human did. Use AI tools to organize or flag candidates, not to make final decisions. Require a recruiter or manager to review any rejection the tool recommends and to record a job-related reason that does not depend on the score alone.

Keep records of which tool was used, what criteria it applied, when it was last tested, and what the results were. If a candidate requests an accommodation or an alternative process, honor it and document the response. For monitoring, publish a clear policy that explains what is collected and how it is used, and obtain acknowledgments.

Questions to ask your agent

Ask whether your EPLI policy defines wrongful employment acts to include invasion of privacy and whether any privacy or biometric statutes are excluded. Ask how the policy treats claims by applicants rather than employees, since screening tools generate applicant claims. Ask whether disparate-impact and class or collective claims are covered, and how the retention applies when many claimants are involved.

Ask whether the policy responds to regulatory investigations into automated decision tools, and whether fines and penalties are excluded. Finally, answer any application questions about AI use accurately; some carriers now ask directly.

Talk with an independent agent

AI hiring and monitoring tools are changing faster than policy forms. An independent agent at Provident Financial Group can review your EPLI policy against the tools you use, quote multiple carriers, and help you compare how each handles privacy and algorithmic-bias allegations. Call (866) 964-6660 or request a quote online.

Frequently asked questions

If a vendor's AI tool screens out applicants unfairly, is the vendor liable instead of us? Generally, the employer that uses the tool to make hiring decisions is the party applicants sue, and employment laws place the responsibility on the employer. Any contractual recourse against the vendor is a separate matter from defending the employment claim itself.

Does EPLI cover a claim that we monitored employees illegally? It depends on the policy. Many EPLI policies include invasion of privacy as a covered wrongful act, which may apply to monitoring claims by employees, but some exclude violations of specific privacy or wiretapping statutes and most exclude statutory penalties. Review the definitions and exclusions with your agent.

Do we have to tell candidates we use AI in hiring? Some jurisdictions require notice, and the trend is toward more disclosure rather than less. Even where it is not required, telling candidates about the tool and offering an alternative on request is a sound practice that supports your defense if a claim is later filed. Confirm current requirements with employment counsel.

Want this reviewed for your business?

Submit once and we'll bring back up to 10 carrier quotes, with the coverage differences explained in plain English. No obligation.

Get Up to 10 Quotes

General information only. Coverage is governed by the terms of the policy actually issued. This article is not legal advice.