Jamie Dimon’s remark in October 2025 comparing fraud to “cockroaches” pushed an often-overlooked corner of finance into public view. Supply chain financing – particularly factoring – is rarely discussed outside specialist circles, yet it underpins the flow of goods behind almost every consumer purchase.
According to Resistant AI, a series of trade finance failures has highlighted just how fragile the system’s trust layer can be, and why technology-driven oversight is becoming urgent.
Factoring has long provided liquidity to sellers by allowing them to turn outstanding invoices into immediate cash. Instead of waiting 30 to 120 days to be paid, the seller offloads the receivable to a factor at a small discount. The factor then collects the full amount when the customer’s payment comes due. It is a straightforward solution, but one that places enormous reliance on the accuracy and authenticity of seller-provided documents.
Over time, as the seller–factor relationship deepens, documentation requirements tend to loosen. What begins with full sets of invoices, purchase orders, and delivery records can gradually devolve into sending little more than an itemised list – and in some cases, only a total figure. This convenience introduces risk, and the industry openly acknowledges that a proportion of financed invoices are “administratively deficient,” a polite euphemism for fake or manipulated documentation.
The collapse of First Brands illustrated how quickly this can spiral. Matt Levine described a pattern of discrepancies where invoices bore little resemblance to actual customer orders, where entirely fictitious invoices were submitted, and where the same receivable was factored multiple times. The fraud typified how easily traditional checks can be bypassed when factors rely on surface-level documentation.
The core issue is simple: checking the invoice itself is rarely enough. Invoices are fully under the seller’s control and therefore easy to fabricate. What matters is validating the commercial substance behind each one. Supporting materials – such as customer purchase orders, bills of lading, and delivery confirmations – provide a more reliable audit trail because they originate from independent third parties. Historically, reviewing these bundles was labour-intensive and often deprioritised. That dynamic is changing.
Advances in AI technology now allow factoring firms to analyse stacks of semi-structured documents rapidly and without human involvement. Resistant AI’s systems can classify uploaded documents, confirm their issuer and origin system, and detect anomalies inconsistent with past behaviour. These tools can spot forged purchase orders, altered dates, mismatched document numbers, or even reused imagery. Legitimate variations, however, can still be recognised, reducing the noise of false alerts that previously burdened compliance teams.
Once the documents are validated, firms can combine extracted data with broader risk indicators. They can identify whether the transaction aligns with historical patterns, whether the customer relationship is established, and whether the amounts correspond to typical order sizes. What emerges is a richer, more accurate picture of the business operations behind each invoice.
This does not eradicate fraud entirely – no system can. But it does make long-running, repetitive schemes significantly harder to sustain. By forcing sellers to submit full documentation packages and applying AI-driven validation, factors can eliminate many of the vulnerabilities that enabled recent failures. Ultimately, AI offers a practical way to illuminate these risk pockets and challenge the outdated assumption that a certain amount of fraud is inevitable.
Copyright © 2026 RegTech Analyst
Copyright © 2018 RegTech Analyst
