- This article forms the fourth chapter of the whitepaper from Trade Finance Global (TFG) and the Bankers Association for Finance and Trade (BAFT): “From lenders to leaders: Banks in flux”.
- Fraud and money laundering are proliferating in a trade environment full of uncertainty.
- As detection tools become more sophisticated, however, so do the crimes.
Financial crime is less glamorous than “an offer he can’t refuse”, or a matter of “his brains or his signature […] on the contract,” as Francis Ford Coppola’s 1972 ‘The Godfather’ details in grim, gory glamour. Nonetheless, the stakes in trade-based fraud and money laundering are a matter of many livelihoods and millions of dollars lost.
Major cases of trade-based fraud have been making headlines in the past year – most recently the First Brands accounting scandal, preceded by the spectacular collapse of supply chain finance startup Stenn in the UK, and a handful of court cases all over the world around forged trade documents. Beyond the headlines, though, SMEs are often the worst affected by financial crime; they are least protected against fraud and lack the means to absorb the losses it creates.
AI has been hailed as the solution, with many hoping it could be a way to protect banks and companies against financial fraud for a fraction of the cost of traditional methods. However, AI’s potential is also becoming clear to would-be criminals, who are increasingly using the tool to trick employees into fraudulent transactions or to falsify documentation. At the same time, innovative AI tools that could immediately spot fraud are often not as effective as human detection, slowing down the fight against financial crime.
Recent headlines brought trade-based fraud to the forefront
In December 2024, the collapse of major British trade finance provider Stenn and the subsequent fraud allegations against its CEO sent shivers down the spine of an industry that is still fresh off the Greensill case 3 years earlier.
In July 2025, the London High Court ruled in favour of Trafigura, the commodities trading giant, in a 2017 case involving forged signatures in tripartite agreements and trade finance documents, bringing the case back to the headlines.
In August 2025, a Singapore ruling against a Swiss bank alleging fraudulent misrepresentations in Letters of Credit further reminded the industry of the prevalence of potential fraud – and the high legal fees associated with remedying it.
The hidden victims
Both directly and indirectly, SMEs suffer most from financial crime. The effects of fraud on SMEs can be far higher than on larger companies, and even the most careful small businesses take a hit – albeit indirectly – when fraud spreads in their market.
As with excessive regulations, a high fraud risk – especially compounded by regulations setting rigorous compliance standards around AML/ KYC – can push banks away from a specific region, market or sector. Even if banks do not completely leave a sector, the first cuts they make to preserve profitability tend to be in the SME loan space, which is less scalable. Applying extensive anti-fraud regulation to SME clients requires more resources for less overall profit than checking the compliance of, say, a multinational corporation that may have established AML monitoring and already carries out KYC checks on suppliers.
Every time there is a rise in trade-based financial crime, regulators tend to impose more requirements for KYC checks on banks, further widening the gap between the high standards set on the trade finance sector and the relatively lighter burden on other areas like payments. The problem of financial crime must be perceived within the industry as a whole, rather than as the sum of the parts which make it, when regulation is being set.
Whenever those high standards lead to a transaction bringing up a red flag – even if it is a minor discrepancy or false alarm – banks have to stop the transaction immediately to investigate further, disrupting the supply chain in both directions. This lengthens timelines for banks, further pushing them away from the trade finance sector as a whole, but it can also be critical for SMEs, who often struggle most to manage working capital and for whom just a few days’ delay of funds could be fatal.
Chaos begets crime
Especially in emerging economies like Africa, fraud shoots up during periods of geopolitical turbulence, with criminals attempting to take advantage of the uncertainty and urgency around situations such as conflicts. Local banks and institutions might be overwhelmed by the changes and less on guard about unusual-looking documents or communications.
“When there’s geopolitical instability, […] we tend to see attempts, usually linked to some sort of syndicate, on things like guarantees and LCs,” said one participant. These attempts usually target smaller, less sophisticated institutions that lack rigorous anti-fraud procedures, such as local or regional banks.
The growing pains of AI
As much as AI is increasing speed and efficiency in many areas of banking, in global trade, it remains, on balance, a liability. While AI is still not at the level of humans in fraud or money-laundering detection, it has an enormous potential to make fraud attempts much more sophisticated and complex. “I’m petrified about the impact that AI is having and how much easier AI is going to make trade finance fraud,” said one participant.
“I’m petrified about the impact that AI is having and how much easier AI is going to make trade finance fraud,” said one participant.
In the past, a big reason why trade finance fraud attempts failed was that they just “felt” wrong – the amounts involved were implausible, the jargon wasn’t used quite correctly, or a document was too perfect to be true. AI is democratising fraud for would-be criminals, removing the technical knowledge that used to be necessary to come up with a credible trade finance fraud or forgery. AI is also making it possible to execute these attempts on a scale, greatly raising the likelihood that one distracted member of a customer service team will fall for it.
While AI is making fraud far easier to execute, it is not doing much to help detection yet. In most banks, nearly all fraud attempts are identified by humans, while automated systems often let them through unflagged. This is because the in-depth knowledge of clients and markets that even low or mid-level employees have is far more valuable than AI’s pattern detection systems: someone who knows a client or region can spot a transaction that doesn’t make sense from miles away, while an LLM or RFI will only look at whether the documents are technically correct and compliant. “I’m less scared, but I’m highly sceptical of it all, really,” said another participant.
“I’m less scared, but I’m highly sceptical of it all, really,” said another participant.
Therefore, a bank’s most valuable resource against fraud remains its talent. Operations employees often have enormous powers to release funds or sign documents in the bank’s name, as well as the skills needed to exercise careful judgment. Beyond continuing fraud detection training for those employees, banks could leverage AI, not to detect fraud itself, but to automate repetitive tasks like drafting or checking documents to leave more time for the back-office staff to fight back against fraud.
