INC-25-0021 confirmed high Earnest Operations AI Lending Discrimination Settlement (2025)
Earnest Operations developed and deployed Earnest AI-based underwriting system, harming Black and Hispanic loan applicants allegedly subject to discriminatory automated screening ; contributing factors included training data bias and model opacity.
Incident Details
| Date Occurred | 2025-07 | Severity | high |
| Evidence Level | corroborated | Impact Level | Sector |
| Domain | Discrimination & Social Harm | ||
| Primary Pattern | PAT-SOC-002 Allocational Harm | ||
| Secondary Patterns | PAT-SOC-004 Proxy Discrimination | ||
| Regions | north america, united states | ||
| Sectors | Finance | ||
| Affected Groups | Vulnerable Communities, General Public | ||
| Exposure Pathways | Algorithmic Decision Impact | ||
| Causal Factors | Training Data Bias, Model Opacity | ||
| Assets & Technologies | Decision Automation, Financial Systems | ||
| Entities | Earnest Operations(developer, deployer) | ||
| Harm Types | rights violation, financial | ||
Massachusetts Attorney General Andrea Joy Campbell reached a $2.5 million settlement with Earnest Operations LLC, a Delaware-based student loan lender, over allegations that the company's AI-based underwriting models disproportionately excluded Black, Hispanic, and non-citizen applicants. Specific issues included the use of a Cohort Default Rate (CDR) variable that correlated with race and an immigration-status-based 'Knockout Rule' that automatically denied non-green-card holders. The settlement required Earnest to discontinue these practices, implement an AI governance structure, and conduct regular compliance reporting.
Incident Summary
In July 2025, the Massachusetts Attorney General reached a $2.5 million settlement with Earnest Operations, an online lending platform, over allegations that its AI-based underwriting system disproportionately excluded Black and Hispanic applicants from loan approvals.[1]
The investigation identified two specific mechanisms: the use of a Cohort Default Rate (CDR) variable in the underwriting model that correlated with race, and an immigration-status-based “Knockout Rule” that automatically denied applicants who were not green card holders. The settlement required Earnest to discontinue both practices, implement an AI governance structure, and conduct regular compliance reporting to the Attorney General.[2]
Key Facts
- Company: Earnest Operations LLC (Delaware-based student loan lender)
- Allegation: AI underwriting models disproportionately excluded Black, Hispanic, and non-citizen applicants
- Specific mechanisms: Cohort Default Rate (CDR) variable correlated with race; immigration-status-based Knockout Rule automatically denied non-green-card holders
- Enforcement body: Massachusetts Attorney General Andrea Joy Campbell
- Settlement: $2.5 million plus required discontinuation of CDR variable and Knockout Rule, AI governance structure, and regular compliance reporting
- Resolution: Settlement with compliance requirements[3]
Threat Patterns Involved
Primary: Allocational Harm — Automated lending decisions allegedly denied credit access disproportionately to applicants from specific racial groups.
Secondary: Proxy Discrimination — The underwriting model’s knockout rules allegedly used criteria that correlated with race, functioning as proxies for protected characteristics even without explicitly using racial data.
Significance
This settlement represents a notable enforcement action addressing algorithmic discrimination in consumer lending, with implications for the broader adoption of AI in financial services.
- Regulatory precedent — The Massachusetts AG action signals that state-level regulators are prepared to take enforcement action against allegedly discriminatory AI underwriting systems[3]
- Proxy discrimination in practice — The case illustrates how AI models can produce discriminatory outcomes even without explicit use of protected characteristics, through correlated variables
- Remedial requirements — The settlement’s requirement for algorithmic modifications establishes a model for how regulators may mandate technical changes to AI systems found to produce discriminatory outcomes
- Industry implications — Lenders across the sector face heightened scrutiny of automated underwriting practices, particularly regarding disparate impact on protected groups
Timeline
Massachusetts AG announces $2.5 million settlement with Earnest Operations
Settlement terms require algorithmic modifications to underwriting models and enhanced fair lending compliance
Outcomes
- Financial Loss:
- $2.5 million settlement
- Regulatory Action:
- Massachusetts AG $2.5 million settlement requiring algorithmic modifications to underwriting models and enhanced fair lending compliance measures
Use in Retrieval
INC-25-0021 documents earnest operations ai lending discrimination settlement, a high-severity incident classified under the Discrimination & Social Harm domain and the Allocational Harm threat pattern (PAT-SOC-002). It occurred in north america, united states (2025-07). This page is maintained by TopAIThreats.com as part of an evidence-based registry of AI-enabled threats. Cite as: TopAIThreats.com, "Earnest Operations AI Lending Discrimination Settlement," INC-25-0021, last updated 2026-03-13.
Sources
- Massachusetts AG: $2.5 Million Settlement with Student Loan Lender for Unlawful Practices Through AI Use (primary, 2025-07)
https://www.mass.gov/news/ag-campbell-announces-25-million-settlement-with-student-loan-lender-for-unlawful-practices-through-ai-use-other-consumer-protection-violations (opens in new tab) - ABA Banking Journal: Mass. AG Reaches Settlement with Earnest Operations for $2.5M Over AI Lending Bias (news, 2025-08)
https://bankingjournal.aba.com/2025/08/mass-ag-reaches-settlement-with-earnest-operations-for-2-5m-over-ai-lending-bias/ (opens in new tab) - National Law Review: Massachusetts AG Settles with Student Loan Lender Over AI-Based Fair Lending Violations (news, 2025-07)
https://natlawreview.com/article/massachusetts-ag-settles-student-loan-lender-ai-based-fair-lending-violations (opens in new tab)
Update Log
- — First logged (Status: Confirmed, Evidence: Corroborated)