Fear the reaper: characterization and fast detection of card skimmers Scaife et al., USENIX Security 2018
Until I can get my hands on a Skim Reaper I’m not sure I’ll ever trust an ATM or other exposed card reading device (e.g., at garages) again!
Scaife et al. conduct a study of skimming devices found by the NYPD Financial Crimes Task Force over a 16 month period. The bad news is that you and I don’t really have much chance of detecting a deployed card skimming device (most of the folk wisdom about how to do so doesn’t really work). The good news is that the Skim Reaper detection device developed in this research project was able to effectively detect 100% of the devices supplied by the NYPD. That’s great if you happen to have a Skim Reader handy to test with before using an ATM. The NYPD are now actively using a set of such devices in the field.
Card skimmers and why they’re so hard for end-users to detect Almost as well-know as (credit and debit) cards themselves is the ease with which fraud can be committed against them. Attackers often acquire card data using skimmers devices attached to legitimate payment terminals that are designed to illicitly capture account information. Once installed, skimmers are nearly invisible to the untrained eye and allow attackers to sell stolen data or create counterfeit cards.A successful card skimming attack reads the magnetic data from your card when you insert it into the slot, and uses a pinhole camera to video you entering your pin on the keypad (or alternatively, a keypad overlay to detect key presses). ATM fraud increased by 40% in 2017, and card skimming fraud in total is projected to reach over $30 billion by 2020.
The authors partnered with the NYPD Financial Crimes Task Force and obtained all of their skimmer devices reports for a 16 month period. This provided information on 35 unique types of skimming device found during the period. ATMs at banks were the most widely attacked.
A minority of skimmers are internally attached (typically at gas stations with weak physical security around the pumps), but most are external devices that can be installed without opening the payment device. External devices can be attached as card reader overlays, deep-inserts inside the magnetic stripe slot, those that fit in the EMV slot (chip reader) and those that wiretap the physical communication line.
The following figure shows an overlay skimmer and how it looks when installed.
Common advice is to ‘pull on the card reader’ or check for anything unusual looking before using an ATM. But the overlays are very cleverly designed to make it very difficult to spot them, and they are attached tightly enough that only a prying tool such as a knife can detect them. And you’re not going to spot a deep-insert skimmer using that technique anyway:
We observe that when law enforcement tries to determine if a payment device has a skimmer, they first look for the camera’s pinhole since it is faster for them to identify than other mechanisms, further indicating that advice such as pulling the card acceptor may not be effective .
Here’s a typical view from a pinhole camera:
The cameras themselves may be cleverly positioned, e.g. in a light making them very hard to see when the light is on:
And then of course there might not be a camera at all, but a PIN pad overlay instead:
There are smartphone apps that claim to detect skimmers by looking for Bluetooth radio communications, but only 20% of the skimmers recovered by the NYPD actually had wireless data retrieval capability, making this defence useless against a large majority of skimmers.
Card skimmers have to read your cardThere’s one thing that’s fundamental to overlay and deep-insert skimmers they have to actually read your card data! This requires a read head pressed against the magnetic track on the card with a spring mechanism. Furthermore, the head must be a conductor and in practice seems to always be metallic. The smallest read head encountered contacted the card over a 1.5mm section of the head, and a lower bound is probably around 1mm.
These three properties constitute fundamental aspects of card reading; that is, we believe that adversaries seeking to read cards reliably must adhere to designs which meet these characteristics. Detecting double reads with the Skim Reaper
If a skimming device is present, then in additional to the legitimate card read there will be a second read by the skimmer. So if we can reliably detect read heads reading from a card, and can count the number of reads, we’d have a way of detecting when more than the expected number of reads happens (‘dip style’ devices that take the card all the way in and then eject it later expect a read on inserting and another one on removal in normal operation).
The Skim Reader comprises a specially designed measurement card to be inserted into card readers, and an attached microcontroller.
The measurement card has a carefully etched set of traces in the magnetic stripe, (aligning with each of the three data tracks). When a read head contacts the card it bridges a pair of electrical traces and completes a circuit back to the microcontroller. Traces are 0.1mm apart, an order of magnitude smaller than the shortest track read length observed in the wild.
The microcontroller applies voltage to one half of the traces and monitors for circuit completion on the opposite half. Once voltage goes above a threshold the controller counts one read and waits for the voltage to drop back below the threshold indicating the read head has fully passed over the card. After this it begins looking for a voltage spike again.
Evaluation
The Skim Reaper is tested with a skimming device installed by the authors in a test bed, and also with 10 real skimmers provided by the NYPD.
Since the team didn’t have access to the devices these real skimmers were designed to be attached to, the test simply involved ensuring that the Skim Reaper could detect the read from the skimming device in isolation. The Skim Reaper was tested 5 times with each device, and successfully detected the skimmers on every attempt.
Skimmers represent a significant and growing threat to payment terminals around the world … we show that robust and portable tools can be developed to help consumers and law enforcement to rapidly detect such attacks.