Document Type
Article
Publication Date
2026
Abstract
In today’s real estate market, more housing decisions are being automated. If artificial intelligence renders a biased decision, will the person impacted have any legal recourse? Commentators have expressed concern that the proprietary nature of AI will allow defendants in discrimination lawsuits to avoid discovery, effectively shielding their decision making from judicial scrutiny. This article examines whether AI and algorithm-related information is the discovery black box that some fear it to be.
Addressing the issue in the context of Fair Housing Act (FHA) litigation, the article argues that concerns about access to AI and algorithm-related discovery are overstated. In real litigation, the Federal Rules of Civil Procedure have proven well-equipped to handle the discovery issues raised by novel AI and algorithm technology. Thus, with proper planning, plaintiffs should be able to obtain the AI and algorithm-related information they need to prove their claims. However, Plaintiffs who file FHA claims against AI developers and users do still face significant challenges – and those challenges are likely to grow. In particular, a circuit split regarding whether FHA plaintiffs can challenge policies where the disparate impact arises from existing demographic patterns makes it difficult for plaintiffs in some circuits to survive motions to dismiss. If the Supreme Court takes up this circuit split, disparate impact liability under the FHA is likely to be significantly narrowed.
While there have been many articles that discuss AI and AI bias, few discuss discovery strategies when litigating claims arising from AI bias. Peeking Inside the Black Box is the first article to comprehensively review the existing case law addressing the discoverability of AI and algorithm related information. It aims to both contribute to the academic debate about discoverability of AI and provide a practical guide for plaintiffs’ attorneys litigating disparate impact claims against AI developers.
