Apple Card algorithm allegedly grants lower credit limits to women
Goldman Sachs faced allegations that its Apple Card algorithm discriminated against women. A regulatory probe by the NY DFS followed, though the regulator eventually found no violation of fair lending laws.
Records by entity: Goldman Sachs
Apple Card's credit limit algorithm was accused of gender bias after women received significantly lower limits than men with similar profiles.
The model's reliance on proxy variables led to gender-based disparities despite the absence of explicit gender data.
Key facts
- What
- Goldman Sachs faced allegations that its Apple Card algorithm discriminated against women.
- Incident date
- Nov 1, 2019
- Who
- Goldman Sachs
- Failure mode
- Policy Violation
- AI surface
- Algorithmic Decision
- Severity
- Medium
What happened
Users alleged that the Apple Card credit limit algorithm provided women with significantly lower credit limits than their husbands despite having similar financial profiles. These claims gained widespread attention after a high profile tech entrepreneur shared his experience on social media. The incident led to an investigation by the New York State Department of Financial Services.
What broke inside the model
The algorithm allegedly relied on data points that acted as proxies for gender even though gender was not an explicit input. This created a disparate impact where the model's automated decisioning process produced biased outcomes. The lack of transparency in the black box model made it difficult for users to understand the reasoning behind their limits.
What it cost
Sources
Cite this entry
https://failureindex.ai/failures/apple-card-algorithm-allegedly-grants-lowerAI Failure Index. "Apple Card algorithm allegedly grants lower credit limits to women" (FI-0335). Realm Labs. https://failureindex.ai/failures/apple-card-algorithm-allegedly-grants-lower (indexed Jun 9, 2026).Data fields CC-BY 4.0, prose citation permitted. Incident ID FI-0335. Full dataset at /data.
Note from Realm Labs, the Index steward
How Realm fits
- Prism
- OmniGuard
This entry sits in the index's predictive wing: a system that scores, ranks, perceives, or steers rather than generates. Realm's runtime layer is built for the generative and agentic systems now moving into these same decision seats, where it watches a model's internal state and holds an unsupported claim or an unchecked action before it commits. The control gap on this record, an automated decision that reached people with no runtime check in front of it, is the same gap. The index keeps predictive failures on the record because the pattern carries straight into the systems shipping today.