Edition: computed from the 56 fintech profiles as of Nov 19, 2026; the rule is stated under Method. The bank equivalent: which banks lead in AI for fraud detection and credit decisioning?
By AI credit-decisioning models in production and disclosed results, the fintechs leading in AI credit decisioning are Affirm, Happen Bank, Plaid, Upstart and Dave: Affirm (3 of 3 in production, led by Transformer-based real-time underwriting); Happen Bank (3 of 3 in production, led by Machine-learning credit decisioning); Plaid (2 of 3 in production, led by Transaction foundation model); Upstart (2 of 2 in production, led by AI personal, auto and HELOC underwriting); Dave (1 of 2 in production, led by CashAI cash-flow underwriting). 21 of the 22 fintechs with an AI credit-decisioning use case on their page run one in production, and 7 disclose an approval, delinquency, accuracy or model-scale figure.
Data as of Nov 19, 2026 · 56 fintechs and neobanks profiled · 22 scored · 52 sources cited below · ranked answer, not a comparison
Which fintechs decide credit with AI, and what do they disclose?
The 22 highest of the 22 fintechs with a credit-underwriting or fair-lending use case that describes a model, AI or automation. Columns: the use cases with their stage, the latest disclosed figure about approvals, delinquency, losses, accuracy or the model itself, the latest dated credit-model move, and the score.
| # | Company | Credit decisioning use cases (stage) | Disclosed result or model figure | Latest credit-model move | In production / score | Status | Date | Deeper | Sources |
|---|---|---|---|---|---|---|---|---|---|
| 1 | Affirm BNPL & lending | Transformer-based real-time underwriting (in production) · Explainable decisions for thin-file applicants (in production) · AdaptAI personalized credit offers (in production) | Monthly-installment 30+ day delinquency rate (ex-Peloton, Pay-in-X): 2.9%, up 35bps QoQ and 9bps YoY (Sep 30, 2026) | Transformer underwriting model goes live (Sep 17, 2026) | 3 of 3 in production · score 10 | In production | Sep 30, 2026 | Affirm: credit decisioning AI | Affirm (Business Wire) ↗ · Affirm Holdings via SEC EDGAR ↗ · Affirm Holdings via SEC EDGAR ↗ |
| 2 | Happen Bank BNPL & lending | Machine-learning credit decisioning (in production) · Automated loan issuance (in production) · Point-of-sale home improvement lending (in production) | Automation rate for issued loans: record-high >90% (Jun 30, 2026) | Q1 2026: over 60 AI initiatives, >90% loan automation; home improvement launch (Apr 27, 2026) | 3 of 3 in production · score 10 | In production | Jul 27, 2026 | profile | LendingClub Corporation via SEC EDGAR ↗ · LendingClub Corporation via PR Newswire ↗ · Happen, Inc. via PR Newswire ↗ |
| 3 | Plaid Business finance & infrastructure | Transaction foundation model (in production) · LendScore 2 and LendScore Arc (in production) · Plaid Income on a transformer-based LLM (rolling out) | LendScore Arc predictive lift (Plaid's claim, early testing): 20% for deep-subprime borrowers; 24% for super-prime borrowers, vs. the core model (Oct 6, 2026) | LendScore 2, LendScore Arc and Instant Link launched (Oct 6, 2026) | 2 of 3 in production · score 9 | In production | Oct 6, 2026 | profile | Plaid ↗ · Plaid ↗ · Plaid ↗ |
| 4 | Upstart BNPL & lending | AI personal, auto and HELOC underwriting (in production) · Alternative-data credit access (in production) | Underwriting model accuracy vs traditional credit scoring: 2.77x, after adding over 140 new features in Q3 2026 (87.27% of the modelled inaccuracy gap remains, per management) (Sep 30, 2026) | Second no-action letter covers the AI underwriting model (Nov 30, 2020) | 2 of 2 in production · score 8 | In production | Sep 30, 2026 | profile | Upstart Holdings via SEC EDGAR ↗ · Consumer Financial Protection Bureau ↗ · The Motley Fool ↗ |
| 5 | Dave Neobank | CashAI cash-flow underwriting (in production) · Dave Flex Card underwriting (pilot) | CashAI v6 impact after full rollout: average ExtraCash size $225 (September 2026); 28-day past-due rate on September originations below 2% (Sep 30, 2026) | CashAI v5.5 fully implemented (Sep 10, 2025) | 1 of 2 in production · score 7 | In production | Nov 5, 2026 | profile | Dave Inc. via SEC EDGAR ↗ · Dave Inc. via GlobeNewswire ↗ · earnings call transcript (The Motley Fool) ↗ |
| 6 | GXS Neobank | Proprietary AI credit models on ecosystem data (in production) · FICO Platform credit decisioning for GXS FlexiLoan (in production) · Lending to thin-file borrowers (in production) | no result or model figure disclosed | GXS AutoLoan for Grab drivers (Sep 9, 2026) | 3 of 3 in production · score 6 | In production | Sep 9, 2026 | profile | GXS Bank ↗ · Thoughtworks (Business Wire) ↗ · FICO (Business Wire) ↗ |
| 7 | Oportun BNPL & lending | Automated credit decisioning (V13) (in production) · Bank-transaction model (Plaid Check) (in production) | no result or model figure disclosed | no dated credit-model move | 2 of 2 in production · score 4 | In production | Aug 12, 2026 | profile | Oportun via SEC EDGAR ↗ · earnings call transcript (The Motley Fool) ↗ · The Financial Brand ↗ |
| 8 | Upgrade Neobank | Gradient-boosting loan decisions (in production) · Real-time fairness monitoring with FairPlay (in production) | no result or model figure disclosed | FairPlay fairness monitoring for ML underwriting (Jan 25, 2024) | 2 of 2 in production · score 4 | In production | Jan 21, 2026 | profile | American Banker ↗ · Forbes ↗ · FairPlay via PR Newswire ↗ |
| 9 | Varo Neobank | Varo Line of Credit decisioning (in production) · Varo Advance small-dollar lending (in production) | no result or model figure disclosed | no dated credit-model move | 2 of 2 in production · score 4 | In production | Feb 2, 2026 | profile | Varo Bank via PR Newswire ↗ · Varo Bank via Business Wire ↗ · PYMNTS ↗ |
| 10 | MariBank Neobank | Transformer-based credit-risk models (Monee) (in production) | Monee credit-model approval uplift: around 10% versus previous models at a similar level of risk (Jun 30, 2026) | Sea details Monee's transformer credit models (Aug 11, 2026) | 1 of 1 in production · score 4 | In production | Sep 23, 2026 | profile | The Motley Fool (earnings call transcript) ↗ · Sea Limited ↗ |
| 11 | OakNorth Neobank | ONCI machine-learning credit intelligence (in production) | Cumulative principal losses: 0.045% over ten years (Dec 31, 2025) | no dated credit-model move | 1 of 1 in production · score 4 | In production | Mar 17, 2026 | profile | OakNorth ↗ · OakNorth ↗ |
| 12 | MoneyLion Neobank | Instacash and credit products (in production) · AI decisioning agent for abandoned applicants (pilot) | no result or model figure disclosed | no dated credit-model move | 1 of 2 in production · score 3 | In production | Oct 29, 2026 | profile | earnings call transcript (The Motley Fool) ↗ · MoneyLion Inc. via SEC EDGAR ↗ · earnings call transcript (The Motley Fool) ↗ |
| 13 | ANEXT Bank Neobank | bettr credit service on Ant's AI Credit Tech (in production) | no result or model figure disclosed | ANEXT lending embedded in AI-powered EPOS360 (Nov 12, 2025) | 1 of 1 in production · score 2 | In production | Nov 12, 2025 | profile | Business Wire ↗ · Ant International ↗ |
| 14 | Block Payments | Machine-learning underwriting (in production) | no result or model figure disclosed | no dated credit-model move | 1 of 1 in production · score 2 | In production | Aug 5, 2026 | Block: credit decisioning AI | Block, Inc. via SEC EDGAR ↗ |
| 15 | Bluevine Neobank | Automated small-business credit decisions (in production) | no result or model figure disclosed | 2026 AI plans: support and faster underwriting (Jan 12, 2026) | 1 of 1 in production · score 2 | In production | Oct 16, 2026 | profile | Valley National Bancorp ↗ · Bluevine ↗ · Bluevine ↗ |
| 16 | Klarna BNPL & lending | Real-time underwriting (in production) | no result or model figure disclosed | no dated credit-model move | 1 of 1 in production · score 2 | In production | Sep 10, 2025 | Klarna: credit decisioning AI | Klarna Group plc via SEC EDGAR ↗ |
| 17 | Monzo Neobank | ML credit decisioning (in production) | no result or model figure disclosed | no dated credit-model move | 1 of 1 in production · score 2 | In production | May 19, 2026 | profile | Monzo ↗ · Monzo ↗ |
| 18 | Nubank Neobank | nuFormer underwriting (in production) | no result or model figure disclosed | Q2 2026: nuFormer in three portfolios, AI agents above 60% of support (Aug 13, 2026) | 1 of 1 in production · score 2 | In production | Aug 13, 2026 | profile | Nu ↗ · American Banker ↗ |
| 19 | Payoneer Business finance & infrastructure | ML underwriting for working capital (in production) | no result or model figure disclosed | no dated credit-model move | 1 of 1 in production · score 2 | In production | Mar 10, 2026 | profile | Payoneer Global Inc. via SEC EDGAR ↗ · Payoneer ↗ · Payoneer ↗ |
| 20 | Stripe Payments | Parafin's Paraformer transformer-based underwriting (in production) | no result or model figure disclosed | Stripe-owned Parafin discloses a transformer-based underwriting model (Oct 23, 2026) | 1 of 1 in production · score 2 | In production | Oct 23, 2026 | profile | American Banker ↗ |
| 21 | Toast Payments | Toast Capital origination models (in production) | no result or model figure disclosed | no dated credit-model move | 1 of 1 in production · score 2 | In production | Oct 15, 2026 | profile | Toast via SEC EDGAR ↗ · Toast ↗ |
| 22 | Chime Neobank | Transformer event-sequence risk models (rolling out) | no result or model figure disclosed | no dated credit-model move | 0 of 1 in production · score 1 | Rolling out | Aug 27, 2026 | profile | Chime ↗ |
Each row quotes the company's page on this site; the company link opens the full record with every source. draft profiles are not counted.
Which fintechs lead in AI credit decisioning within each category?
The same ranking read within the site's fintech categories: up to four companies per category, with their rank in the table above.
| Category | Rank | Company | Use cases in production / described |
|---|---|---|---|
| Neobank | 5 | Dave | 1 of 2 in production: CashAI cash-flow underwriting · Dave Flex Card underwriting |
| Neobank | 6 | GXS | 3 of 3 in production: Proprietary AI credit models on ecosystem data · FICO Platform credit decisioning for GXS FlexiLoan |
| Neobank | 8 | Upgrade | 2 of 2 in production: Gradient-boosting loan decisions · Real-time fairness monitoring with FairPlay |
| Neobank | 9 | Varo | 2 of 2 in production: Varo Line of Credit decisioning · Varo Advance small-dollar lending |
| BNPL & lending | 1 | Affirm | 3 of 3 in production: Transformer-based real-time underwriting · Explainable decisions for thin-file applicants |
| BNPL & lending | 2 | Happen Bank | 3 of 3 in production: Machine-learning credit decisioning · Automated loan issuance |
| BNPL & lending | 4 | Upstart | 2 of 2 in production: AI personal, auto and HELOC underwriting · Alternative-data credit access |
| BNPL & lending | 7 | Oportun | 2 of 2 in production: Automated credit decisioning (V13) · Bank-transaction model (Plaid Check) |
| Payments | 14 | Block | 1 of 1 in production: Machine-learning underwriting |
| Payments | 20 | Stripe | 1 of 1 in production: Parafin's Paraformer transformer-based underwriting |
| Payments | 21 | Toast | 1 of 1 in production: Toast Capital origination models |
| Business finance & infrastructure | 3 | Plaid | 2 of 3 in production: Transaction foundation model · LendScore 2 and LendScore Arc |
| Business finance & infrastructure | 19 | Payoneer | 1 of 1 in production: ML underwriting for working capital |
How is this ranking computed?
Data: the 56 published fintech profiles, as of Nov 19, 2026. Use cases are the ones each page tags credit-underwriting or fair-lending, kept only when their own text describes a model, AI, machine learning or automation; credit decided by a partner bank or lender, and products with no credit decision (a no-credit-check card), are left out. A figure is a disclosed number about approvals, defaults, delinquency, past-due rates, credit losses, predictive power or accuracy, automation, or the model's own scale (variables, repayment events, loans in its data); revenue, origination totals, loan books and loss allowances are left out.
Score = 2 × AI credit-decisioning use cases in production + 1 × those at other stages + 2 × disclosed figures (capped at 2). Ties: use cases in production, then figures, then name. 22 companies qualify; the top 22 are shown.
What are the caveats?
Most rows are lenders deciding their own credit, but a row can describe a credit-risk score or model that a company builds for other lenders to decide with; the use-case column names each system. Model-scale figures (variables, repayment events) describe the model, not its results.
Credit decisions are under fair-lending and adverse-action rules, so lenders describe their models cautiously and results are self-reported against each lender's own baseline. A company missing from the table has no qualifying entry on its page on this site; that is a gap in its public record as recorded here, not proof that it does none of this work.
The bank ranking on the nearest question, which banks lead in AI for fraud detection and credit decisioning?, is computed from the bank profiles with its own rule; the peer groups and their sources differ, so positions are not comparable across the two pages.
Which fintech lender leads in AI credit decisioning?
Affirm: 3 of 3 AI credit-decisioning use cases in production, led by Transformer-based real-time underwriting and Explainable decisions for thin-file applicants; its page reports Monthly-installment 30+ day delinquency rate (ex-Peloton, Pay-in-X): 2.9%, up 35bps QoQ and 9bps YoY (Sep 30, 2026).
Which fintechs disclose results from AI underwriting?
Affirm (Monthly-installment 30+ day delinquency rate (ex-Peloton, Pay-in-X): 2.9%, up 35bps QoQ…), Happen Bank (Automation rate for issued loans: record-high >90%), Plaid (LendScore Arc predictive lift (Plaid's claim, early testing): 20% for deep-subprime…), Upstart (Underwriting model accuracy vs traditional credit scoring: 2.77x, after adding over 140…), Dave (CashAI v6 impact after full rollout: average ExtraCash size $225 (September 2026)…) and MariBank (Monee credit-model approval uplift: around 10% versus previous models at a similar level…).
Which neobanks use AI to decide credit?
11 neobanks run an AI credit model in production: Dave (CashAI cash-flow underwriting), GXS (Proprietary AI credit models on ecosystem data), Upgrade (Gradient-boosting loan decisions), Varo (Varo Line of Credit decisioning), MariBank (Transformer-based credit-risk models (Monee)), OakNorth (ONCI machine-learning credit intelligence), MoneyLion (Instacash and credit products) and ANEXT Bank (bettr credit service on Ant's AI Credit Tech).
How many fintechs use AI for credit decisions?
21 of the 56 profiled fintechs describe an AI credit-decisioning model in production; 22 describe one at some stage.
What rules apply to AI credit decisions?
In the US, ECOA and Regulation B adverse-action notices and fair-lending law apply however a decision is made; in the EU, the AI Act treats creditworthiness scoring as high-risk. The bank fraud and credit ranking and the AI credit underwriting guide cover the rules.
When any of these companies moves on AI, you'll read it the next morning.
when a bank, fintech or regulator moves on AI, it is in the next morning's brief · six sourced stories · 7 am PT · free
plus every regulation, bank and agent page update, the morning after · leave any morning