Massachusetts Settles AI Bias Case with Student Loan Firm

According to a Sheppard Mullin article by A.J. Dhaliwal, Mehul Madia & Beineng Zhang, Massachusetts Attorney General Andrea Joy Campbell announced a $2.5 million settlement with a student loan company over alleged AI bias violations of the Massachusetts Consumer Protection Act and the Equal Credit Opportunity Act.

The enforcement action, Commonwealth of Massachusetts Superior Court Civil Action In The Matter Of Earnest Operations LLC, was resolved through an assurance of discontinuance in Suffolk County Superior Court.

It centered on claims that the company’s underwriting systems, including artificial intelligence-driven models, produced disparate impacts on protected groups and lacked sufficient oversight.

The investigation examined both automated and manual processes used for issuing and refinancing student loans.

Regulators alleged that the company failed to test its AI models for fair lending risks, used inputs that disadvantaged non-white and non-citizen applicants, and allowed unregulated manual overrides.

According to the settlement, school-level default data that factored into underwriting decisions negatively affected certain racial groups, while adverse action notices often omitted the true reasons for denials due to system design flaws.

Additionally, the company allegedly operated without adequate policies, testing protocols, or documentation to ensure compliance with fair lending obligations.

The settlement terms require the company to establish AI governance procedures, conduct annual fair lending tests, remove specific underwriting variables, and submit compliance reports to the Attorney General’s Office.

It also mandates enhanced oversight of underwriting processes to prevent discriminatory effects and ensure consistent application of credit decisions.

The agreement pairs a monetary penalty with injunctive relief, signaling a robust approach to addressing alleged disparate impact.

The case serves as a significant example of state-level legal action targeting algorithmic bias and governance gaps in financial services.

The settlement terms may inform compliance strategies for entities using AI or complex underwriting models, particularly in jurisdictions pursuing aggressive fair lending enforcement. 

Counsel advising such organizations should evaluate the use of variables that may have disproportionate impacts, implement structured override controls, and establish ongoing monitoring and testing to reduce exposure to similar enforcement risks.