A network of 23 shell companies domiciled in Zimbabwe and South Africa has been implicated in siphoning US$450 million through illicit financial flows (IFFs), part of a looting scourge draining a combined US$3 billion annually from the two neighbours.
This is according to a ground-breaking study in the Journal of Risk and Financial Management, which analysed 1,8 million transactions using data from South Africa’s Financial Intelligence Centre (FIC), the Reserve Bank of Zimbabwe (RBZ) and SWIFT. The research deployed an Artificial Intelligence tool called FALCON to mine transaction records and expose cross-border laundering patterns with unprecedented precision.
FALCON achieved 98,7% accuracy, according to the 18-page report, “Disruption in Southern Africa’s Money Laundering Activity by Artificial Intelligence (AI) Technologies”. It revealed the depth of illicit flows between Africa’s most industrialised economy and its economically strained northern neighbour.
For Zimbabwe, it is the latest in a string of reports showing the extent of economic haemorrhage caused by porous borders. A Zimbabwe National Chamber of Commerce study last year identified smuggling hotspots as drivers of shocking job losses, with an estimated 18 000 positions wiped out in just a few years.
From gold to diamonds and lithium, Zimbabwe’s authorities have struggled to stem losses that economist Eddie Cross this week said were draining potential state revenues into the hands of shadowy operators.
“The rise in illicit financial activities across the South Africa–Zimbabwe corridor, with an estimated annual loss of US$3,1 billion demands advanced AI solutions to augment traditional detection methods,” the paper said.
It noted the South Africa–Zimbabwe corridor is plagued by systemic volatility in anti-money laundering (AML) frameworks.
“By leveraging data from South Africa’s FIC, Zimbabwe’s RBZ, and SWIFT, FALCON achieved 98,7%, surpassing Random Forest (72,1%) and human auditors (64,5%), while reducing false positives to 1,2%,” it said.
- Mavhunga puts DeMbare into Chibuku quarterfinals
- Bulls to charge into Zimbabwe gold stocks
- Ndiraya concerned as goals dry up
- Letters: How solar power is transforming African farms
Keep Reading
The AI model, developed at India’s National Forensic Sciences University, meets both FATF and other compliance standards, with 92% judicial admissibility. It can process two million transactions per second at a cost of US$0,002 per 1 000 transactions, making it a cost effective option for emerging markets.
“Using FALCON, US$450 million in concealed funds were detected from 23 shell companies in South Africa and Zimbabwe, with 94,0% cross-border precision, achieving 94% cross-border detection precision” the report said.
The biggest laundering channels are trade-based money laundering, especially gold export mis-invoicing, and cryptocurrency layering — both used to disguise smuggling proceeds. These exploit regulatory gaps, cash-intensive transactions and fragmented enforcement between the two countries.
It claimed Zimbabwe’s official systems fail to detect 42% of cross-border laundering linked to mis-invoiced trade and cash-based transactions. This means large volumes of smuggling proceeds — including illegal gold shipments — are likely moving undetected through formal and informal channels.
Weaknesses include over-reliance on rule-based reporting, which criminals circumvent, and the absence of integrated analysis of both transaction patterns and networks of linked companies.
Gold smuggling, fuel smuggling and under-invoicing of imports and exports fit squarely into the trade-based laundering typology outlined in the paper.
These schemes often involve shell companies in both Zimbabwe and South Africa to move value without triggering red flags. The laundering process typically features rapid cross-border transfers, sometimes within hours, routed through multiple jurisdictions to obscure origins, according to the report.
FALCON’s hybrid AI architecture combines transformer models and graph neural networks to track suspicious transaction sequences and map linked entities in real time.
In trials using RBZ and FIC data, it outperformed older models and human auditors, offering a major potential boost to Zimbabwe’s AML arsenal, the report showed.
For the RBZ, adopting such tools could sharply improve detection of smuggling-linked flows in gold, fuel and other high-risk sectors. The model can generate court-ready evidence and could be integrated into other systems to block illicit arbitrage.
The study emphasises that joint South Africa–Zimbabwe data sharing is critical — the model’s effectiveness depends on both countries supplying real transaction and company registry data. AI-driven systems can also reduce false positives to 1,2%, cutting wasted investigative effort and costs, experts said.
The findings land as Zimbabwe battles a surge in financial crimes. FIU data shows that between 2019 and 2024, smuggling accounted for US$920 million, illegal gold and precious stone dealings US$880 million, corruption US$730 million, fraud US$500 million, tax evasion US$300 million, and drug trafficking US$170 million.
Total illicit proceeds over the five-year span may have reached US$6,15 billion — roughly US$1,23 billion per year.
The RBZ’s 2024 Financial Stability Report also exposed large-scale IFFs in real estate and motor vehicle dealerships, revealing industrial-scale looting beyond the minerals sector. It said laundering, tax evasion and smuggling are thriving in under-regulated sectors where cash dominates and AML controls are weak.
“Real estate, car dealers and precious stone or precious metal dealers are the sectors that are most susceptible to money laundering,” the RBZ said.
