PODCAST | The many-headed hydra of financial crime

Listen to this podcast on Spotify, Apple Podcasts, Podbean, Podtail, ListenNotes, TuneIn

The growth of AI in document forgery, the lethargic response of legislation, and the complexity of international commodity transactions all pose significant challenges to combating financial crime. 

Charles Osborne, Director at Trade Finance Global (TFG), sat down with Jonas Rey, CEO of Athena Intelligence, to break down the issues of fraud vulnerability, the impact of digitisation and AI, and how to defend against financial crime. 

What drives large-scale fraud? 

In a few short but memorable months in 2020, two traders based in Singapore were found to have committed fraudulent activity, affecting major banks and rattling the global commodities industry for years after. 

Agritrade International Pte Ltd collapsed following the company’s deception of around two dozen lenders, resulting in nearly $470 million in bank losses. In the same year, Hin Leong also defaulted after it emerged the company had deceived HSBC into providing almost $112 million in financing for fictitious oil trades. 

The COVID pandemic challenged commodities markets across the world, with oil prices going negative for the first time in history in May 2020. Both frauds turned on liquidity access, meaning they could be sustained only as long as the money in commodities kept flowing.. Agritrade and Hin Leong employed fraudulent methods, including falsifying accounts and pledging the same Bill of Lading to multiple banks, which allowed them to access significant financing without the collateral and trades to back it up.  

Rey suggested that government incentives encouraging banks to turn Singapore into the capital of commodity finance led to a rapid influx of new money, creating an environment that facilitated the fraud.

Rey explained that banks were incentivised to loan large amounts to Singapore start-ups and traders, but “you can’t deploy capital fast on a large scale while maintaining good controls and a due diligence process.”

The sheer size of the frauds made 2020 a turning point. The fallout forced banks to implement various due diligence, monitoring, and compliance processes over the following five years to avoid a repeat. 

The problem with a bad reputation

Since 2020, banks have tightened their due diligence processes. However, this has led to an increasing exclusion of SMEs, which frequently lack the necessary resources to meet administrative standards, preventing them from securing financing. 

Rey suggested that banks now often rely on the reputation of an industry, business, commodity, or region to assess risk rather than the nuance of each individual company. 

“Truth has somehow become secondary to the reputation side of the business,” said Rey. Whole industries and regions can become “tarnished” by a high-profile fraud case, even if their operation was completely independent. 

A similar trend towards perception-based judgment is happening in regulating compliance with Russian sanctions. 

Commodities traders tend to be “a bit too legalistic when it comes to sanction compliance,” said Rey, explaining that the current UK and EU sanctions regime is more of a “political” than a “legal process.” 

For SMEs, the problem is two-fold. They often lack the resources to meet the administrative requirements for accessing financing and navigating sanctions and tariff markets. Yet they also lack the resources to manage their image, making it challenging to maintain institutional access in a reputation-driven era.

The post-2020 shift towards perception and reputation as key values for accessing financing and facilitating sanctions compliance could pose a significant risk to the industry, exacerbating the trade finance gap. 

Slow trials?

Trade finance, in particular, is challenging to prosecute due to the global nature of commodity shipments, which often involve multiple international jurisdictions and sets of laws. 

Following Balli Steel’s collapse in early 2013, with a shortfall of $500 million, the UK Serious Fraud Office (SFO) initiated an investigation into the alleged fraud. Still, the former CEO and his colleagues were not found guilty and sentenced until April 2023, a decade later. Only about $600,000 was recovered, less than 2% of the total losses.

Rey, who worked on the Balli Steel case, pointed out that the length of time taken to prosecute meant the legislation did not act as a deterrent. 

With even well-funded organisations like the SFO taking a decade to bring a case to conviction, reaching conviction stages can be almost impossible for countries with fewer resources.  

“I think the issue is not the commodity sector, but more that the legal world and the investigative powers are not yet adapted to trade in a business or an industry that is, in its essence, global,” Rey explained.

In September 2025, a new “failure to prevent fraud” law came into force in the UK, designed to facilitate the prosecution of large companies.

Introduced in the Economic Crime and Corporate Transparency Act (ECCTA) in 2023, the law means companies can face prosecution if an associated person commits a fraud offence for the benefit of the company, amending the legislation to avoid focusing solely on company directors. 

The ECCTA also introduced laws impacting all-sized companies. Under section 196, the SFO can prosecute companies of any size if a “senior manager” commits any of a long list of offences, including forgery, money laundering, and false accounting. 

Is digitisation set to help or hinder the fight against financial crime?

Using AI for real-time anomaly detection in cross-border transactions has been hailed as a solution to combat fraud. 

Swift recently highlighted the success of its experiment, which utilised Privacy Enhancing Technologies (PETs) to share fraud insights across 13 banks securely, enabling it to train AI locally while upholding privacy laws. 

However, blockchain solutions are only as good as the data they use. LLMs perpetuate any mistakes in the original data set, and the larger the network, the greater the impact this can have. 

Blockchain and digital solutions also often fail to capture a crucial aspect of commodities trading. As Rey points out, we sometimes tend to forget that trading involves real goods that can be destroyed, tampered with, altered, or stolen.   

AI also has the potential to increase the capacity of people to commit fraud. 

“I think everything that is digital can be faked,” Rey said. AI can convincingly and instantly alter documents, increasing the ease with which individuals can conduct fraud.

The rise of AI-generated deepfakes, which mimic the voices, speech, and behaviour of individuals, also poses new problems. This often renders the advice that banks used to follow to avoid scams and fraud – check everything with a phone call, look for unusual tone or vocabulary in emails – obsolete.

 However, this too can be an easy fix, said Rey.  Taking the time to slow down before responding and double-checking with other people first could be the best defence against pressure from AI scammers. 

Slaying the many-headed hydra

As Rey reminds us, “fraud will never go away. It’s an illusion to believe that you can totally root it out.”

Instead, mitigating financial crime is about minimising the risk and scale and improving the speed at which institutions can respond and tackle the problem. 

In the era of instant payments and AI, transactions have become much more complicated, particularly for trade finance. This is exacerbated by the industry’s inherent need to navigate a multitude of international jurisdictions, geopolitical tensions and sanctions, a range of commodities and cooperations, and a complex web of physical and digital documents. 

Fighting a many-headed hydra requires a many-headed defence, fighting each head at its root.