$24 billions check fraud

I began my career with the U.S. Secret Service (USSS) in 1998 and was assigned to the New Haven, CT, Resident Office. For those who are unaware of the history of the USSS, the agency was founded in 1865 in the U.S. Department of the Treasury in order to combat a massive amount of counterfeit currency circulating in the country at the close of the U.S. Civil War. It wasn’t until 1901, after the murders of three U.S. Presidents that the USSS was given the responsibility of protecting the President of the United States. When the USSS was founded, the only other Federal law enforcement agencies in existence were the U.S. Marshal Service and the U.S. Postal Inspection Service (USPIS). It wasn’t until 1908 that the Federal Bureau of Investigation (FBI) was founded. After the terrorist attacks on 9/11, the Department of Homeland Security (DHS) was created and the USSS was reassigned from Treasury to DHS. Yes, the government has so many acronyms but I promise not to use anymore and there will not be a test at the end. Even the USSS was reassigned to DHS, the USSS not only maintained its core violations such as investigating counterfeiting and other white collar crimes, the USA PATRIOT ACT charged the USSS with establishing Electronic Crimes Task Forces, now called Cyber Fraud Task Forces all around the country with the purpose of investigating sophisticated cyber crimes. Yes, there are acronyms for those Task Forces too but a promise made is a promise kept. As such, the USSS has deep roots in protecting the financial integrity of the United States and has been investigating all forms of white collar crime for the last 159 years.

Now you know the history of the USSS, it will make sense as to why the first several cases assigned to me and just about every agent in those days were called “check cases.” Check cases were either stolen checks that were forged and cashed, counterfeit corporate checks, altered checks in which the dollar amounts were changed, or what we called deceased payee check cases. The latter referred to government checks issued, typically monthly, to someone receiving some form of government assistance but the “payee” died and the family didn’t notify the government so as to continue receiving and cashing those monthly checks. Check fraud was so rampant in the 90s that USSS Field Offices had entire squads of agents assigned to investigate these types of cases. Those squads were called, to nobody’s surprise, check fraud squads. 

Cases used to get assigned to us newbie agents in yellow envelopes containing some of the information receives from either a bank of the Treasury outlining the basic facts of the case. Our job was to investigate and bring charges to those responsible thru either local courts or the U.S. Attorney’s Office. I so badly want to add in acronyms. 

Working cases in those days were a lot easier than they are today. The tools in our investigative toolbox for check cases included reviewing surveillance photos, interviewing tellers and witnesses, and talking to the local detectives in the area to see if they recognized the suspects in the surveillance photos. Once we identified suspects, we conducted physical surveillance, might set up a “pole camera” to remotely conduct surveillance, did trash pulls (it is what it sounds like…going thru their trash after they disposed of it), creating photo arrays with the suspect and 7 other “mopes” who looked the suspect to show to witnesses for identification, sending the checks to the lab to see if there are any latent fingerprints, and last but not least, requiring the suspect, sometimes thru court order, to provide handwriting exemplars to send to the lab for comparison. Handwriting exemplars meant the suspect would write words or signatures a whole bunch of times so the lab analysts could compare them with the ”bad” check. 

Those cases were fun to work and you learned a lot from the mistakes you made in interviewing or your investigative process. We were also lucky in that we worked regularly with the USPIS when the checks were stolen from the mail. Like the USSS, the USPIS is chalk full of bloodhound investigators that could run circles around many of the larger agencies who might or might not have entire blocks of TV programming dedicated to them. Our little office probably arrested 30-40 check “forgers” from 1998-2003. 

Over time, criminals moved away from check fraud for other more lucrative financial crimes that didn’t require physically showing up at a bank or check cashing place and risk getting caught on surveillance cameras. The ease in which criminals could purchase and re-encode stolen credit cards became a very attractive alternative with higher yields and less risk. Criminals, like you, have their own KRIs and KPIs to keep the money flowing while staying out of jail. 

A few years ago, check fraud came back with a reckless abandon. Criminals realized they could steal checks en mass right out of the mail due to some vulnerabilities in the system such as a generic key, called an arrow key, that can open up those blue mailboxes on the street corner and clustered mailboxes all over the country. Steal one arrow key and the mail system is your oyster. So now its easier to get stolen checks and cheap software makes it very easy to alter them without much effort. And finally and most importantly, they can deposit those checks right from their burner phone without fear of getting physically caught on surveillance cameras or leaving fingerprints on the checks. No bueno for the good guys in law enforcement and the banking sector. 

Remember all those investigative techniques I mentioned earlier? Think about how many of them are now out of the equation due to the new manner in which fraudsters can deposit and cash out these checks. Coupled with the ability to take advantage of the “float time” for banks to deposit and provide the funds to the account holder who deposited the check, it has been a field day for the bad guys. 

That said, new technology for the bad guys can also mean deploying new technology for the good guys. Handwriting analysis was normally a key component to the successful conviction of check fraudsters back in the day. Although the days of handwriting exemplars are largely over, new technology has replaced it. I’m happy to report the new technology is so much more effective in detecting …my thought is this is where I turn it over to you guys to talk about haystack and how it packs all the investigative steps mentioned into one solution…… 

As traditional investigative methods become less effective, leveraging AI technology offers a promising solution to the complexities of check fraud detection. The key challenge lies in accurately verifying payee information amid diverse handwriting styles and check designs. To address this, AI models for detecting check fraud should undergo continuous training on a vast array of handwriting and check formats, enabling them to extract and compare handwritten details with the provided payee and amount data and subsequently deliver a reliability score for the match. 

A recent breakthrough was achieved using a state-of-the-art AI to identify handwritten check payee details. This AI system employs a blend of Convolutional Neural Network (CNN), Bi-Directional Long-Short-Term Memory (LSTM), and Connectionist Temporal Classification (CTC) technologies. These components work together to analyze varying dimensions and layouts of the check fields, facilitating accurate character recognition in the handwritten sections and generating a confidence level regarding the match with the check's metadata. Additionally, incorporating investigator feedback into the AI's training process further refines the model's accuracy. 

The rise of generative AI will likely make check fraud more complex, urging businesses to adopt safer transaction methods and banks to invest in AI for large-scale fraud detection to counteract this evolving threat. 

Previous
Previous

Streaming A.I. voice vishing

Next
Next

GoodLabs Studio will receive close to a million dollars in funding from FedDev Ontario to commercialize quantum liquidity optimizer