1. Huskey v. State Farm Fire & Casualty Co. (2022–ongoing)

Core allegation

Black homeowners allege that State Farm used algorithmic claim-processing systems that flagged Black policyholders’ property insurance claims for greater scrutiny and delay than similarly situated white policyholders.

What the algorithm allegedly did

• Used predictive models in claims triage and fraud detection
• Flagged certain claims as “high risk”
• Routed those claims into longer investigation workflows
• Resulted in delayed payouts for repairs and recovery

Legal claims

• Violation of the Fair Housing Act
• Disparate impact discrimination (race-neutral model producing racial disparity)
• Unfair claims handling practices

Court’s key ruling point

The federal court allowed the case to proceed, finding the plaintiffs plausibly alleged:

• Statistical disparities in claim treatment
• A connection between algorithmic processes and racial outcomes

This is crucial because it means:

The court accepted that algorithmic discrimination is legally plausible, even without full access to the model.

Why it matters

• One of the clearest U.S. insurance AI bias class actions surviving early dismissal
• Forces discovery into model logic, training data, and fraud detection rules

2. Estate of Lokken v. UnitedHealth Group (nH Predict litigation)

Core allegation

The estate of a Medicare Advantage patient alleges that UnitedHealth Group used an AI system (often referenced as nH Predict) to deny or prematurely cut post-acute care coverage.

How the algorithm allegedly worked

• Predicted “expected length of stay” or recovery trajectory
• Recommended discharge or denial of continued coverage
• Overrides or heavily influences physician decisions

Legal issues raised

• Breach of insurance contract obligations
• Bad faith denial of benefits
• Whether reliance on algorithmic prediction violates Medicare Advantage coverage rules

Procedural significance

• Currently in discovery disputes
• Plaintiffs are seeking access to:
  • model architecture
  • decision thresholds
  • internal validation metrics

Why this case is important

It represents a shift from “bias in pricing” to:

“AI as a direct gatekeeper of life-saving medical insurance benefits.”

This case is often cited as a prototype for AI accountability in health insurance denial systems.

3. SafeRent tenant screening algorithm litigation (Louis v. SafeRent Solutions)

Why it matters for insurance law analogies

Although technically housing, courts treat it as directly relevant because it involves insurance-like risk scoring systems applied to protected classes.

Core allegation

Plaintiffs claimed SafeRent Solutions used an algorithmic scoring system that:

• Assigned lower scores to Black and Hispanic renters
• Penalized applicants using housing vouchers
• Had no meaningful explanation or appeal system

Outcome

• Case settled for $2.3 million
• Company agreed to limit or suspend use of scoring model for years

Legal principle established

Even if:

• inputs are “neutral”
• model is statistical

If outcomes systematically harm protected groups → Fair Housing Act liability risk exists

Why it matters for insurance bias litigation

Insurance courts often borrow this reasoning:

If algorithmic scoring determines access/pricing and disproportionately impacts protected classes, it can trigger discrimination liability even without intent.

4. EEOC v. iTutorGroup (AI hiring discrimination settlement)

Why it matters (insurance analogy case)

Although not insurance, it is foundational for algorithmic bias law.

Allegation

iTutorGroup used automated hiring software that:

• Automatically rejected older applicants
• Used age-based filtering rules embedded in the algorithm

Legal claims

• Age discrimination under the Age Discrimination in Employment Act (ADEA)
• Disparate treatment via automated decision system

Outcome

• $365,000 settlement
• Injunctive relief requiring changes to automated screening

Legal significance

This is one of the first formal regulatory enforcement actions involving algorithmic discrimination

Why insurance lawyers care

Insurance underwriting systems often use:

• age proxies
• risk scoring models
• automated eligibility filters

So courts use this case to support the idea that:

Algorithmic automation does NOT reduce legal responsibility for discrimination.

5. Mobley v. Workday (AI discrimination platform liability case)

Core allegation

A plaintiff sued Workday alleging its AI hiring tools systematically rejected:

• Black applicants
• older applicants
• disabled applicants

Key legal innovation

The court allowed claims to proceed on the theory that:

The AI vendor itself can be treated as an “agent” participating in discrimination.

Why this matters for insurance AI

Insurance companies increasingly rely on vendors for:

• underwriting models
• fraud detection systems
• claims scoring engines

This case suggests:

Vendors and insurers may BOTH be liable for biased algorithmic outputs.

Legal impact

• Expands liability beyond insurer → includes algorithm providers
• Breaks the “we didn’t design it, we just used it” defense

6. State Farm class action (racial bias in claims algorithms)

Core allegation

A class of Black homeowners alleges State Farm used AI tools that:

• subjected their claims to heightened scrutiny
• delayed payouts for repairs
• caused unequal treatment compared to white policyholders

Legal framing

• Fair Housing Act violations
• Disparate impact via claims automation

Key judicial reasoning

A federal judge allowed parts of the case to proceed, emphasizing:

• statistical disparity evidence was enough at early stage
• plaintiffs did not need full algorithm disclosure initially

Why it matters

It reinforces a key modern rule:

Plaintiffs can survive dismissal even when algorithms are opaque, as long as statistical bias is plausible.

Big Picture: What these cases collectively establish

Across insurance algorithm bias litigation, courts are converging on five legal principles:

1. Disparate impact applies to AI

Even neutral algorithms can be illegal if outcomes disproportionately harm protected groups.

2. “Black box” is not a defense

Insurers cannot avoid liability by claiming proprietary model secrecy.

3. Discovery pressure is critical

Courts are increasingly forcing disclosure of:

• training data
• model logic
• risk scoring thresholds

4. Vendors can be liable

AI providers may share responsibility with insurers.

5. Claims handling algorithms are the highest risk area

Compared to pricing models, claims denial systems face the strongest litigation exposure.