Sardine launched Sardine AI Labs, an applied research group focused on fraud and financial crime AI.
What happened
- Sardine opened applications for up to five research fellowships for independent researchers.
- Sardine reports its card foundation model improved fraud detection accuracy by 68% for a consumer card issuer and 41% for a business card issuer versus standard machine-learning approaches, in testing with issuers excluded from training.
- The card model was trained on about one billion transactions over two years from over a dozen card issuers.
- Applications for the AI Fellowship Program close December 13, 2026, with the inaugural fellows announced mid-January 2027.
Why it matters to Signicat
- Sardine's ReadID-adjacent fraud claims are not relevant to Signicat's core eID brokering; the card model targets transaction fraud, a different workflow from Signicat's identity-verification and authentication lines.
- The announcement is a research and talent program with vendor-reported accuracy gains; no independent validation or customer deployment is stated, so the impact on Signicat's competitive position is low-to-moderate.
Evidence from the source (6 quotes)
- “Sardine is opening applications for up to 5 research fellowships” source ↗
- “In testing with issuers that were entirely excluded from the model’s training data, the model improved fraud detection accuracy by 68% for a consumer card issuer and 41% for a business card issuer compared with standard machine-learning approaches.” source ↗
- “The card model was trained on about a billion transactions over the last 2 years from over a dozen card issuers.” source ↗
- “Applications are open through December 13, 2026.” source ↗
- “The team will study how AI can learn the language of financial behavior, anticipate new attacks and help risk teams make better decisions.” source ↗
- “The model architecture, training methodology, and full results are available in a new technical whitepaper.” source ↗
Fraud deepfake injectionAML KYB screeningThought leadershipResearch reportHiring signalGlobal
5 of 5 facts verified against the source · confidence high · claude-haiku-5-5