Harper v. Sirius XM: The Real Implications Of Using AI Hiring Tools

This post follows up on last week’s discussion of the use of artificial intelligence in hiring.

As artificial intelligence becomes a fixture in the recruiting and hiring process, employers who rely on algorithmic screening tools are facing a new wave of scrutiny — and a pending case out of the Eastern District of Michigan is one to watch closely.

In Harper v. Sirius XM Radio, LLC, No. 2:25-cv-12403 (E.D. Mich.), plaintiff Arshon Harper alleges that Sirius XM’s use of its AI application screening tool (called, “iCIMS”) discriminated against him on the basis of race. Harper claims he applied to roughly 150 positions with Sirius XM over several years and was rejected for each one despite arguably being qualified for many of them.

Harper’s complaint alleges that the iCIMS platform scores and ranks applicants using data points, such as the schools they attended, their home zip codes, and their employment histories, that can function as proxies for race. Harper contends this scoring methodology, built on historical hiring data, effectively imports decades of human hiring bias into an automated system and then applies that bias at scale, disproportionately screening out African American applicants.

Harper brings claims under both Title VII of the Civil Rights Act of 1964 and Section 1981, meaning he asserts both a disparate-impact theory (the tool produces discriminatory outcomes regardless of intent) and an intentional-discrimination theory. He also seeks to proceed as a class action on behalf of similarly situated applicants and asks the court for injunctive relief that would require Sirius XM to stop using, or substantially modify, its AI-driven screening process, in addition to compensatory and punitive damages.

It’s worth emphasizing that these are, at this stage, only allegations. Sirius XM has not yet had its claims tested on the merits, and the court has made no factual findings.

Why This Case Matters

Harper is part of a fast-growing category of litigation testing how existing anti-discrimination law applies to AI-driven hiring decisions. A few features of this case make it particularly important for employers to watch. First, the defendant is the employer and not the AI technology vendor, meaning employers cannot contract out or bypass their Title VII liability simply by outsourcing the technology used to make screening and hiring decisions. Next, Harper alleges that ostensibly neutral inputs, like zip code, school, and work history, correlate to race closely enough that using them at scale produces a discriminatory result. Because this is being applied to a machine-learning system, it puts added responsibility on employers to understand not just what their AI tools can do, but how they do it. Lastly, its worth noting that Harper brought this case pro se – that is, representing himself without an attorney. As awareness of algorithmic hiring bias spreads, employers should expect more claims like this one, including from unrepresented plaintiffs, and should not assume that only sophisticated plaintiff’s firms will bring them.

Looking Ahead

Harper v. Sirius XM is still in its early stages, and how the court handles class certification, the sufficiency of the disparate-impact allegations, and any dispositive motions will shape how much practical guidance the case ultimately offers. But even at the pleading stage, it’s a clear signal that AI-driven hiring tools are now squarely within the crosshairs of employment discrimination litigation. Employers who have not yet audited their use of these tools should treat this case as a reason to do so now, rather than after they’ve been served with a complaint of their own.