AI vs Strategic Freight Crime
Cargo theft has existed since the beginning of time & with the evolution of the internet in the past few decades, freight crime evolved into its own digital deception landscape.
A recent report by CNBC showcased how vulnerable U.S. supply chains are becoming due to the growing number of stolen cargo, digitally, with ID loopholes being used to socially engineer & spoof load boards.
Generally speaking scammers have a basic understanding of the logistics system they are targeting, knowing that speed is both the core factor, but also the main vulnerability of the supply chain. Depending on the load board parameters, fraudsters fake their documents based on the standards & templates used by stakeholders (combined with stolen credentials) and then proceed onto using various penetration methods to spoof transactions between carriers & brokers.
It sounds too simple to be true but experts estimate the ballpark loss cumulating to about $1 billion due to freight fraud last year, with most items targeted being either high-value products like consumer electronics or things that are difficult to trace, such as food.
Due to the fractured nature of the industry with little regulation enforcement (even for verification standards) the load board systems are built with mainly a focus on speed (not security) which opens a can of vulnerabilities for organized crime groups to exploit transactions - using techniques like phishing (e.g. fake emails representing a real broker/trucker), social engineering (stealing MC numbers, impersonating dispatchers etc.), scraping data (stealing information from data banks e.g. FMCSA's website) & using other methods such as physical coordination tactics to evade the crime scene on the same day a load was stolen/delivered.
These thefts spiral into causing even more additional chaos for the supply chain as truckers unknowingly involved with fraud are usually left unpaid for their deliveries, brokers loose reputation, while insurers keep raising premiums.
You may think that you're good at spotting phishing emails..,but in fast paced environments like freight brokerages where office staff manage thousands of emails per day, forged domains & branding can often fool a tired person.
One way AI can help combat this is with LLMs (trained on detecting phishing data from labeled datasets) that parse through emails to isolate those with malicious links. Those solutions can be as simple as workflows (like N8N) & plugins that scan each incoming email to assist teams dealing with high volume inboxes.
The challenge however is not in coming up with an AI solution, but to test your application & make sure it works. It involves understanding your infrastructure well enough to make sure the pattern recognition algorithm is the right fit for your data.
If a load board or TMS has only a few fraudulent incidents & not enough metadata for an AI to learn something about it, it may start profiling behavior that may be uncommon but not necessarily fraudulent.
A recurrent theme that rises from this discussion is "Open Source" - will the freight industry be forced to share some data in order to combat fraud in the age of AI? That's a tough one for competition & privacy issues...is AI enough of a 'solution'? If it is, how are those incidents going to get logged to track behavior?
Will we have to come up with some sort of blockchain or AI combined smart contract to keep those logs from being tampered with or lost?
Logically, federally regulated industries will have to develop some sort of AI intelligene gathering network to combat fraud on all levels (or i guess quantum computing).
From my observation - carriers, shippers & brokers that adopt AI with due diligence asap will benefit in operational efficiency. As the market evolves & legacy digital environments become more optimizable, those that will find ways to process information faster & more structurally will have a better market advantage.