Massachusetts AG settled with Earnest for $2.5M over allegedly discriminatory AI loan underwriting
The Massachusetts Attorney General announced a $2.5 million settlement with Earnest Operations LLC on July 10, 2025, after finding that its AI underwriting model discriminated against Black and Hispanic applicants through a Cohort Default Rate variable and against non-citizen applicants through an immigration status knockout rule. Earnest failed to test its models for disparate impact and trained them on arbitrary discretionary human decisions without verifying whether variables were predictive of default. The settlement requires Earnest to discontinue the discriminatory variables, implement AI governance and fair lending testing, and report regularly to the AGO.
A Cohort Default Rate variable and an immigration status knockout rule baked proxy discrimination into Earnest's AI underwriting model, penalizing Black and Hispanic applicants while automatically denying non-citizen borrowers.
Key facts
- What
- The Massachusetts Attorney General announced a $2.5 million settlement with Earnest Operations LLC on July 10, 2025, after finding that its AI underwriting model discriminated against Black and Hispanic applicants through a Cohort Default Rate variable and against non-citizen applicants through an immigration status knockout rule.
- Incident date
- Jul 10, 2025
- Who
- Earnest Operations LLC
- Failure mode
- Policy Violation
- AI surface
- Agentic Workflow
- Severity
- High
What happened
Earnest Operations LLC used AI algorithmic models to make student loan eligibility and pricing decisions that included a Cohort Default Rate variable and an immigration status knockout rule. The Cohort Default Rate variable penalized applicants from schools with higher default rates, disproportionately harming Black and Hispanic applicants, while the knockout rule automatically denied applicants without a green card, creating disparate outcomes based on national origin. The Massachusetts AG found that Earnest failed to test its models for disparate impact, trained them on arbitrary discretionary human selections, and sent inaccurate adverse action notices. On July 10, 2025, the AG announced a $2.5 million settlement requiring Earnest to discontinue the discriminatory variables, implement AI governance and fair lending compliance, and regularly report to the AGO.
What broke inside the model
- 01 · TriggerA prompt pushes against a deployment boundary.
- 02 · Model stepThe model produces the disallowed output.
- 03 · Control gapNo enforcement blocks it at generation time.
- 04 · FailureThe output crosses the policy line.
- 05 · ConsequenceA limit the business set is breached in public.
The output crosses a policy boundary the deployment had defined.
The Cohort Default Rate variable used school-level default averages as a proxy that correlated with race, penalizing Black and Hispanic applicants more than White applicants in approval rates and loan terms. The immigration status knockout rule automatically denied applicants who lacked a green card, creating disparate outcomes based on national origin. Earnest failed to test its models for disparate impact and trained them on arbitrary discretionary human selections without verifying whether the variables were actually predictive of default.
What it cost
Sources
- PrimaryAG Campbell Announces $2.5 Million Settlement With Student Loan Lender For Unlawful Practices Through AI Use, Other Consumer Protection Violationsmass.gov
- PressMassachusetts AG Settles Fair Lending Action Based Upon AI Underwriting Modelcfsreview.com
- PressMass. AG reaches settlement with student loan firm for $2.5M over AI lending biasbankingjournal.aba.com
Cite this entry
https://failureindex.ai/failures/massachusetts-ag-settled-earnest-2-5mAI Failure Index. "Massachusetts AG settled with Earnest for $2.5M over allegedly discriminatory AI loan underwriting" (FI-0083). Realm Labs. https://failureindex.ai/failures/massachusetts-ag-settled-earnest-2-5m (indexed Jun 4, 2026).Data fields CC-BY 4.0, prose citation permitted. Incident ID FI-0083. Full dataset at /data.
Note from Realm Labs, the Index steward
How Realm would have caught this
- Prism
- OmniGuard
Realm compares what the model is about to output or do against the policy that governs the deployment, in real time, and can deny or redact the action before it takes effect, which is the gap an after-the-fact review never closes in time.