The news
Plaid announced on October 6, 2026, a set of credit, fraud and payment risk models built on cash flow data, led by Plaid LendScore 2, as part of its annual fall product release. The company says the tools help lenders and payment firms make better decisions, including for borrowers that traditional credit scores miss.
LendScore 2 is a credit risk score that uses cash flow underwriting, which judges a borrower by the money moving through bank accounts rather than only by credit history. Plaid says it predicts a borrower's ability to repay with 42% greater strength than traditional credit data alone, according to PYMNTS. Plaid also introduced LendScore Arc, which it calls its first transformer-based credit risk score, using the same type of AI architecture behind large language models.
Plaid released versions tuned for specific loans. According to Open Banking Expo, Plaid said an auto-lending version lowered delinquency by 26% among deep-subprime applicants and a home-lending version approved 6.3% more borrowers at the same level of risk. Plaid said LendScore Arc showed a 20% lift on deep-subprime and a 24% lift on superprime borrowers. These are Plaid's own test results.
A new feature, Instant Link, lets consumers share cash flow insights with lenders in seconds through Plaid's consumer reporting agency, PYMNTS reported. On fraud, Plaid said a new AI foundation model, a large general model trained on broad data and then adapted to tasks, now powers its Plaid Protect product and delivered up to 40% relative improvement over previous baselines in internal evaluations.
For payments, a sequential model that reads an account's transaction history in order now powers Signal, Plaid's ACH payment risk model. ACH is the US network for bank transfers. Plaid said the model helped Signal prevent 26% more ACH returns, failed or reversed transfers, without increasing false flags in testing. Chief Technology Officer Will Robinson said the models bring deep financial context to every problem they solve, as reported by Open Banking Expo.
The numbers
- LendScore 2 repayment prediction vs traditional credit data (Plaid)
- 42% stronger
- Auto version: delinquency cut, deep subprime (Plaid test)
- 26%
- Home version: extra approvals at same risk (Plaid test)
- 6.3%
- Fraud model improvement over baselines (internal)
- up to 40%
- Additional ACH returns prevented by Signal (testing)
- 26%
Why CEOs should care
For chief credit officers and lending CEOs, cash flow scores can widen the pool of approvable borrowers, especially thin-file applicants with little credit history. But every headline number here is from Plaid's own testing. Before buying, run the model on your own historical applications, compare it with your current scorecard, and ask how it performs across income bands and protected groups. Fair-lending rules still apply, and regulators will expect you to explain adverse decisions.
For CFOs and payment leaders at companies that pull money by ACH, such as subscription businesses, lenders and marketplaces, fewer returns means fewer failed collections and lower fees. Ask for the false-positive rate on your traffic, how the model handles new accounts, and how it helps with Nacha's new ACH fraud monitoring rules; PYMNTS cited a survey finding 94% of firms lack full compliance readiness for them.
For CISOs and data officers, these models depend on consumer-permissioned bank data. Review consent flows, data retention and how Plaid's consumer reporting agency role changes your obligations under credit reporting law.
The bigger picture
Cash flow underwriting has been gaining ground as lenders look for signals beyond FICO-style scores. Plaid, which connects apps to bank accounts, is using the data flowing through its network to sell decision tools, not just connections. That puts it in competition with credit bureaus and specialist fraud vendors.
The use of transformer and foundation models in credit is a notable step. Such models can be harder to explain than traditional scorecards, which matters when lenders must give reasons for declining a borrower.
What’s next
Watch for lender adoption announcements, independent validation of Plaid's performance claims, and how regulators treat cash flow and AI-based scores in fair-lending reviews.
What “Fact-checked” means
Fact-checking means testing a story’s facts against the evidence before it is published. This story went through at least two separate checks before this version was published.
- What we checked
- Its names, figures, dates, job titles, quotes and who said what were checked against the story’s sources, including its main source where it could be opened. The headline was checked for accuracy and overstatement.
- How
- A first check reviewed the whole story. If it passed, a second, skeptical check went back to the sources to look for mistakes in the most important facts. If a check flagged the story, it was edited to fix the problems found, and a separate re-check then reviewed the whole story again.
- Who
- The checks are made by our newsroom, as steps kept separate from the writing, under rules set by our editor, Hussein Mukhtar. A story the checks still flag is held for the editor, who decides whether it is fixed, published or dropped.
- If something is wrong
- “Fact-checked” does not mean error-free. If a material error is found after publication, we correct the story and add a note saying what changed. Report an error
