I just checked an account recently used at Target, and it had about $800 in fraudulent charges made in the last couple days.

Called the bank, cancelled the card. This is becoming almost routine, the third breach of my personal accounts in two years.

I wonder why learning algorithms aren't preventing more fraud. My card has a very clear pattern of use -- locations, merchants, etc. The fraudulent charges are invariably in distant cities, at merchants I've never patronized.

Yet occasionally the bank will take the initiative and shut off an account for suspected fraud -- for charges made within 2 miles of my home, and merchants I've done business with for years, etc.

With petabytes of transaction and fraud data, why can't they build more intelligent fraud detection?