Flagged transactions only, sorted by descending risk. Filters stay out of this path.
| ID | Evidence | Card | Risk | Open |
|---|---|---|---|---|
| tx_000959 | AliExpress 11 small online transactions on this card within 60 minutes. | card_049 | 70 high | Open |
| tx_000950 | AliExpress 8 small online transactions on this card within 60 minutes. | card_042 | 69 high | Open |
| tx_001007 | QuickPay Online Amount is 28.8x this card's median transaction. | card_040 | 67 high | Open |
| tx_000975 | AliExpress 10 small online transactions on this card within 60 minutes. | card_038 | 67 high | Open |
| tx_000981 | AliExpress Amount is 34.9x this card's median transaction. | card_048 | 67 high | Open |
| tx_000965 | AliExpress 11 small online transactions on this card within 60 minutes. | card_049 | 66 high | Open |
| tx_000699 | Videotron First observed use of this device for the card. | card_037 | 66 medium | Open |
| tx_000990 | Apple Store High-value electronics transaction for CAD 1,158.32. | card_020 | 66 medium | Open |
New detector flags and review decisions land here as the working audit stream.
Red is reserved for high-risk review pressure.
Patterns are pulled from card baselines, velocity, category surprise, graph reuse, and model consensus.
Current threshold is 49. The review rate is 8%. Increasing reviewer cost should shrink the queue; increasing missed-fraud cost should widen it.
The command center is organized around the judged product path.