1. State v. Loomis (Wisconsin Supreme Court, 2016)

Area: Criminal sentencing / risk assessment algorithms

Facts:

Eric Loomis was sentenced using a risk assessment tool called COMPAS (Correctional Offender Management Profiling for Alternative Sanctions), which predicted his likelihood of reoffending. The defendant argued that the algorithm was:

• Proprietary (no transparency)
• Potentially racially biased
• Not independently verifiable

Legal Issue:

Whether using a proprietary algorithm in sentencing violates due process rights.

Holding:

The Wisconsin Supreme Court allowed use of COMPAS but imposed strict limitations:

• Judges must be warned about algorithmic limitations
• Risk scores cannot be the sole basis for sentencing
• The tool’s proprietary nature limits transparency, raising fairness concerns

Significance:

This is one of the earliest landmark cases recognizing that algorithmic tools can introduce due process and fairness risks in judicial decisions.

2. Mobley v. Workday, Inc. (N.D. California, ongoing since 2024)

Area: Employment discrimination / AI hiring tools

Facts:

Plaintiff alleged that Workday’s AI-driven hiring screening tools discriminated against applicants based on:

• Race
• Age
• Disability

The system allegedly “automatically filtered out” candidates through algorithmic scoring and resume screening.

Legal Issue:

Whether an AI vendor (not just employer) can be liable under Title VII and ADEA (Age Discrimination in Employment Act) for discriminatory outcomes caused by hiring algorithms.

Court Ruling (procedural stage):

The federal court allowed key claims to proceed, recognizing:

• AI vendors can potentially be treated as “agents” of employers
• Algorithmic decision systems may trigger liability if they function as gatekeepers in hiring

Significance:

This case is foundational because it directly targets AI vendors, not just employers, expanding liability for algorithmic bias in employment systems.

3. EEOC v. iTutorGroup, Inc. (2023 Settlement)

Area: Employment / age discrimination via AI screening

Facts:

The Equal Employment Opportunity Commission (EEOC) alleged that iTutorGroup used automated hiring software that:

• Rejected female applicants over age 55
• Rejected male applicants over age 60
• Did so automatically without human review

Legal Issue:

Violation of Age Discrimination in Employment Act (ADEA) through algorithmic screening tools.

Outcome:

The company settled and paid monetary damages without admitting wrongdoing.

Significance:

This was one of the first major U.S. federal enforcement actions confirming that automated hiring systems must comply with anti-discrimination laws just like human decision-makers.

4. Patel v. Facebook, Inc. (9th Circuit, 2019–2022)

Area: Facial recognition / biometric data / algorithmic profiling

Facts:

Users alleged Facebook’s facial recognition technology collected and stored biometric identifiers without proper consent, violating Illinois Biometric Information Privacy Act (BIPA).

Legal Issue:

Whether automated facial recognition systems unlawfully collect biometric data and enable discriminatory profiling risks.

Holding:

The court allowed the class action to proceed, rejecting arguments that plaintiffs lacked standing.

Eventually, Facebook settled for hundreds of millions of dollars.

Significance:

Although framed as privacy litigation, the case is critical for algorithmic bias concerns because facial recognition systems have been widely criticized for racial and gender misidentification disparities.

5. In re Facebook Biometric Information Privacy Litigation (N.D. Illinois)

Area: Facial recognition / class action algorithmic surveillance

Facts:

Multiple lawsuits consolidated alleging Facebook:

• Used facial recognition without informed consent
• Built biometric templates from users’ photos
• Created potential for discriminatory profiling and surveillance misuse

Legal Issue:

Violation of Illinois BIPA and whether algorithmic biometric systems create unlawful data structures.

Outcome:

Facebook agreed to a major settlement (hundreds of millions of dollars).

Significance:

This litigation helped establish that automated biometric systems are subject to strict consent and fairness requirements, reinforcing accountability for algorithmic systems.

6. U.S. Department of Housing and Urban Development (HUD) v. Facebook (2022 Settlement)

Area: Housing discrimination / targeted advertising algorithms

Facts:

HUD alleged that Facebook’s advertising algorithm allowed landlords and advertisers to:

• Exclude users from seeing housing ads based on protected characteristics
• Use algorithmic targeting that effectively enabled redlining

Legal Issue:

Violation of the Fair Housing Act (FHA) through algorithmic ad delivery systems.

Outcome:

Facebook agreed to:

• Overhaul its housing ad targeting systems
• Remove certain discriminatory targeting tools
• Pay penalties and comply with monitoring requirements

Significance:

This is a landmark case showing that algorithmic ad delivery systems can constitute discriminatory housing practices, even without explicit human intent.

Overall Legal Themes in Algorithmic Bias Litigation

Across these cases, U.S. courts and regulators are converging on several principles:

1. No “AI exception” to civil rights law

Algorithms are treated as extensions of human decision-making systems.

2. Liability can extend beyond users to developers

Cases like Mobley v. Workday suggest AI vendors may share responsibility.

3. Transparency is a growing legal issue

Black-box algorithms raise due process and evidentiary concerns.

4. Disparate impact theory remains central

Even without intent, discriminatory outcomes can trigger liability.

5. Regulation is evolving through litigation

Because federal AI-specific laws are limited, courts are adapting existing statutes.