The Future of Financial Crime

In this Sybrin Game Changers conversation, Nic Swingler, Group Head of Financial Crime at Absa, tells Colin Iles why compliance has shifted from a tick-box exercise to a genuine risk discipline, why South Africa's low fine count doesn't mean low crime, and why bank exits, not fines, are often the real enforcement mechanism. Swingler also explains why proof of address disappeared from onboarding, why state capture went undetected for so long, and how Absa is deploying AI in live, human-reviewed production.

From tick-box to risk based: what actually changed in compliance

Swingler frames financial crime as any crime whose proceeds are moved, kept, or laundered through the financial system, and says his job is to detect, report, and mitigate that exposure, never to eliminate it. Before running Absa's Financial Crime Unit he was chief operating officer for the corporate and investment bank, which gave him technology, product, and data experience he says matters more here than a law degree. He describes the old model as ticking boxes to comply with an act, which he calls "a guaranteed recipe for failure," versus a risk based approach that asks what the firm is actually managing, not just which fields were collected. The shift isn't finished: he calls it "an ongoing battle that every organization, especially the larger ones, will fight for the foreseeable future."

The proof of address example: what deregulation actually looks like

A concrete case shows what moving away from prescriptive rules means in practice. Absa used to require proof of residential address for onboarding, written into the original Financial Intelligence Centre Act, but dropped it after the Act was amended roughly three years before this conversation, because address showed no real correlation to a customer's risk. Swingler says the bank couldn't justify treating a Johannesburg resident as higher risk than someone in Bloemfontein or Durban, so the field was cut except for politically exposed persons, where address still helps map associates and connections. In business banking, the same review cut onboarding data fields by roughly 70%, after the team found many fields existed only because "the next person who takes on the job just does it because the previous person did it."

Why South Africa's fine numbers are low but the crime isn't

Swingler pushes back on reading South Africa's small number of financial crime fines, next to multi-billion-dollar penalties against Goldman Sachs, Wells Fargo, JPMorgan, Westpac, and Citi, as evidence the country has less financial crime. He points to state capture and rhino horn smuggling as clear evidence crime is moving through the banking system, and says banks are largely discharging their reporting obligations; the shortfall is in what happens after a suspicious transaction report reaches the authorities. He cites a recent FATF mutual evaluation, finalised about two months earlier, that found deficiencies across South Africa's end-to-end financial crime chain, meaning enforcement gaps sit with law enforcement and prosecutors more than with the banks' own reporting. He also frames bank exits, quietly closing a client's accounts when reputational risk outweighs the relationship, as a real enforcement lever that never shows up in fine statistics.

Same risk, same rules: the fintech and crypto competitive problem

Swingler agrees with the argument, publicly made by JPMorgan's Jamie Dimon, that banks carry a competitive disadvantage against fintechs and crypto providers that pose bank-equivalent risk without bank-equivalent regulatory burden. A fintech can onboard a customer in minutes without a banking licence, while Absa must meet the full weight of the Banking Act, FIC requirements, and its own board-level risk appetite. His fix is "same risk, same rules": if a fintech or crypto provider poses the same risk as a bank, it should face the same obligations, though he says this convergence isn't moving as fast as he'd like. On crypto, he says the asset itself is neutral, "a credit card doesn't misbehave, it's the person who uses the card who misbehaves," but banks can't verify the provenance of funds entering from unregulated crypto exchanges, citing a case where a fraud victim's funds moved into a crypto exchange under two deceased people's identities and were gone within 20 minutes.

Collaboration and technology: UBO registers, SAMLIT, and AI in production

Absa has pushed ultimate beneficial ownership analysis "to the warm body" with its correspondent and clearing banks for roughly a decade, ahead of regulation, and South Africa's new law now requires the same depth from all banks; Swingler expects a national beneficial ownership register within 12 to 18 months. A new banking-industry workgroup, including the Information Regulator and the FIC, is examining legal barriers to banks sharing information with each other, while SAMLIT, the South African Anti-Money Laundering Integrated Task Force, already gives banks a safe channel to share targeted information with the FIC. On technology, Swingler says self-sovereign identity is a clear step up from physical ID documents for fraud resistance and customer experience, and confirms Absa runs AI models in live production for financial crime decisioning, but only in human-reviewed, fully audited form with no autonomous, self-learning deployments.

Jump to a section of the interview

All links open the full recording on YouTube.

• Heading up a financial crime team is a huge job
• Is financial crime monitoring a tick-box exercise?
• Is there much financial crime in Africa (and where are the $bill fines)
• Why was state capture not picked up?
• Would it help if the banks could collaborate?
• Is crypto a game changer for the banks or the criminals?

Why was South African state capture not picked up sooner by the banks?

Swingler says banks fulfilled their reporting role, submitting thousands of suspicious transaction reports, but the FATF evaluation and the Zondo Commission both raised questions about what happened to those reports afterward, suggesting capture of other institutions may have slowed action on them. He notes Absa was the first bank to publicly exit Gupta-linked companies, which prompted other banks to follow, and says bribery and corruption are especially hard to detect because the underlying transaction can look like an ordinary commercial payment, with no way for a bank to see an inflated invoice without stepping outside its role.

Is crypto a net benefit or a net risk for the banks?

Swingler says the crypto asset itself isn't the problem, the risk sits in the quality of onboarding and the traceability of funds entering an exchange before they reach a bank. Absa customers are free to buy, sell, and invest in crypto, but the bank may decline to do business with specific crypto providers whose risk profile it can't verify, and he expects "same risk, same rules" to eventually extend formal regulation, likely under the FIC, to crypto providers.

Is Absa using AI to detect financial crime today?

Yes, but only in supervised, human-reviewed form. Swingler says Absa has AI elements in live production, built to reflect the decisions its own analysts would make, tested and quality-assured before deployment, with a full audit trail for every decision the model informs. He says the bank hasn't deployed unsupervised, self-learning models and doesn't expect to during his working lifetime.