Entersekt Warns AI Fraud is Outpacing SA Bank Defences

In Short

• AI fraud in South Africa is increasingly using deepfake voice calls that sound like real bank staff.
• A global anti-fraud benchmark report says only 7% of organisations feel more than moderately prepared.
• Experts warn consumer liability rises when victims approve transactions themselves.

What Happened

Entersekt, a financial authentication company, is warning that AI fraud is moving faster than many South African banks can respond. A common pattern is a phone call from an AI agent that sounds like a bank employee. The agent uses personal details, like an ID number, to build trust.

Entersekt fraud analyst Nishan Maharaj says victims are being coached in real time to approve transactions and share credentials. Social engineering means tricking people into doing something unsafe, like giving away a one-time PIN, similar to convincing someone to unlock the door for a stranger.

The Association of Certified Fraud Examiners 2026 Anti-Fraud Technology Benchmarking Report, which includes input from Sub-Saharan Africa, lists the AI-driven schemes cited as rising most over the last two years. Deepfake social engineering was cited by 44% of respondents. Consumer fraud and scams was cited by 38%.

Looking ahead two years, respondents expect strong growth in genAI document fraud and forgery at 55%. They also expect deepfake social engineering at 55%, and deepfake digital injection at 54%. Deepfake digital injection is when fake video, voice, or data is inserted into a verification process, like slipping a forged ID into a checkpoint.

Why It Matters

The report suggests most organisations are not ready for AI-enabled attacks. Only 7% say they are more than moderately prepared to detect and prevent AI-powered fraud.

Maharaj says the “safety net” many consumers expect is weaker than before. If a customer is manipulated into approving a payment or handing over login details, the bank may have limited controls to stop it. That shifts risk toward the user, and raises the need for clearer fraud education, faster incident response, and better visibility across banking systems.

For banks, the challenge is also economic and operational. Maharaj argues legacy technology, slow change processes, and limited data visibility can make prevention harder. He also warns that today’s manageable losses can scale quickly once criminals find a repeatable weakness.