Imagine a million AI agents trying to hack your systems
If you think artificial intelligence (AI), cryptocurrency and quantum computing are rocking your world, then spare a thought for your colleagues and peers trying to stay ahead of the game in the fast-paced world of anti-money laundering, fraud and financial crime. Hot on the heels of the 2026 Insurance Crime Bureau Conference, your writer found himself in the mix at the nCino KYC Africa Fighting Financial Crime event.
The great convergence
The presentation that really hit home was delivered by Rianné Potgieter, CEO at the International Federation of Compliance Associations (IFCA), who spoke about the great convergence underway on both sides of the anti-money laundering and countering the financing of terrorism (AML-CFT) battle lines. Compliance professionals and enforcement agencies are working feverishly to adopt and understand emerging technologies to defend against criminal syndicates that are already deploying them at scale.
“These technologies have a multiplier effect,” Potgieter said. “We cannot just prepare for AI or for crypto or for quantum; we must understand the interplay between these concepts and disciplines and how they impact in our sphere.” One of the major challenges is that the country’s large banks and insurers are moving large legacy systems into the digital world with tight regulatory oversight, whereas crypto-native threat actors can deploy the latest technology available, from scratch and with no regard for the law.
Your writer took a moment to come up with an analogy, landing on the Russia-Ukraine war. This four-year-old conflict illustrates how a dominant military force, with far more jets, ships and tanks, is increasingly being pushed back by a combination of AI and drones. Over recent months, the frontlines in this conflict have moved away from the dirty trenches as drone pilots seek and destroy targets deep within enemy territory. Drones are cheap and expendable; and human pilots can now oversee a series of coded instructions to get the drone to the area it is needed in, set it to search for a target and confirm that target for destruction.
Imagine getting scammed by your kid
To get this newsletter back on track, your writer stumbled upon a recent Moneyweb.co.za podcast published under the title, ‘Imagine getting scammed by your own son’. It speaks of a world where threat actors can use a combination of generative AI tools to create video and voice assets that are increasingly difficult to tell apart from the real thing. Why should compliance teams at brokers and insurers care?
As the presenter explained, the synthetic AI profiles being used by syndicates to open bank accounts or buy insurance are now so good that they can fool all but the most up-to-date fraud detection systems. And it gets worse. Criminal syndicates are opening synthetic AI bank accounts with the sole purpose of making automated push payments for money deposited by catfishing or phishing victims.
Money in these accounts is automatically swept into the cryptocurrency world to purchase dollar- or rand-based stablecoins, from where it can be mixed and layered, or ‘disappeared’ for want of a better word. “It is already extremely difficult for us to identify beneficial owners (BOs) with our traditional ways of looking at AML, and it may not be possible once the money has gone through those processes,” Potgieter said.
The fascinating observation here is that if a criminal compromises your firms know-your-client (KYC) protocols at onboarding, it introduces a type of Trojan horse dilemma. Other departments across your firm will treat the account or policy on the basis that it could not exist had it not gone through onboarding and cleared the BO and KYC hurdles. In this context, it has become non-negotiable to break the silos that exist in larger firms, and have someone who is responsible for overseeing end-to-end compliance and fraud responses.
Financial crime at scale
The next warning was that the combination of AI agents, being code that handles a specific role or function, and agentic AI, being code that plays a supervisory role over thousands of AI agents, will accelerate financial crime into a new realm. According to Potgieter, firms’ systems may face synchronised attacks involving millions of bots. “Where it finds vulnerabilities, it will enter; where it is blocked, it will continue to change, iterate and learn until it can break in,” she said. “That is what is coming our way from a fraud perspective.”
As the AML-CFT experts in the audience sank deeper into their chairs, the presenter dropped the Q-day bomb. There are growing concerns that quantum computing, probably 10 years from commercial viability, is going to render current data encryption moot. Criminals are already stockpiling today’s encrypted data in the hope of ‘mining’ it later. “A decade from now there are going to be quantum computers that will make today’s encrypted data look like child’s play,” Potgieter said. Data protection will have to evolve from encryption to post-quantum cryptography to prevent a future ‘run’ on bank accounts and connected crypto wallets.
The appropriate response to rising fraud and anti-money laundering (FRAML) risk is to move beyond siloed financial crime controls. Safeguards should be put in place across the teams responsible for client onboarding, compliance, IT security and risk management, among others, to ensure that things that look okay at a transactional level are flagged from an end-to-end view. At a practical level, attendees were encouraged to develop a FRAML culture supported by clear accountability, joined-up monitoring and shared intelligence.
Converged strategies for FRAML prevention
“Our strategies in AI and crypto and FRAML have to converge … and all of the role players within an organisation must understand this converged strategy in their own context,” Potgieter said. More broadly, firms will have to move from inflexible, rules-based controls to adaptive and intelligence-led risk management. This, plus keeping human judgement in the loop, is the only way to stay ahead of the growing cohort of criminals using AI-generated identities and crypto rails to break through legacy safeguards.
The presentation closed on an ominous note, with an explainer of once-science-fiction constructs such as artificial general intelligence (AGI), defined as AI with human-level skills across all disciplines, and artificial superintelligence (ASI), where AI exceeds human intelligence. Potgieter said ASI was a future issue; but felt AGI was possible within five years. She concluded that the AI agents being deployed in FRAML prevention were also working for criminals, introducing the threat of machine-speed financial crime in a your-AI versus my-AI world.
Writer’s thoughts:
A future dominated by AI agents at the beck-and-call of agentic AI is sending fear through the ordinary rank and file, especially in areas like tackling fraud and anti-money laundering. Do you think financial and risk advice practices have the financial resources and skills needed to respond to the threat? Please comment below, interact with us on X at @fanews_online or email us your thoughts [email protected].
