AI deepfake fraud surges across Africa as firms race to stop new scams

Businesses across Africa are facing a sharp rise in AI-powered fraud as cybercriminals increasingly deploy deepfake technology to bypass traditional security systems, pushing companies in the banking, fintech, and crypto sectors to rethink how they fight digital crime.

A new Identity Fraud Report by Sumsub revealed that Tanzania recorded Africa’s highest fraud rate in 2025 at 5.0 percent, while Uganda followed at 4.7 percent. Fraud cases in Côte d’Ivoire rose 51 percent year-on-year to 4.5 percent.

In Kenya, deepfake-related attacks now account for nearly 10 percent of fraud attempts, despite an overall decline in fraud activity. Meanwhile, South Africa saw deepfake incidents jump by more than 269 percent year-on-year even as total fraud rates fell by 31 percent.

The figures highlight how cybercriminals are moving away from basic scams toward more sophisticated AI-enabled impersonation attacks capable of deceiving both users and security systems.

Read also: Nigeria joins 60 global regulators to crack down on AI-generated deepfakes

Industry analysts say the development poses a growing threat to Africa’s rapidly expanding digital economy, where financial services, online payments, mobile banking and crypto transactions are becoming more widespread.

To tackle this issue, Sumsub has launched a new fraud prevention tool called Adaptive Deepfake Detector, designed to identify and respond to emerging deepfake scams in real time instead of relying on periodic software updates.

Unlike conventional fraud detection tools that may require weeks or months to update against new threats, Sumsub said its new system uses machine learning to continuously study fraud patterns and adapt within hours.

The company explained that the platform analyses multiple data layers simultaneously, including facial biometrics, device information, IP addresses, geolocation, documents and behavioural signals, to detect suspicious activity before fraud is completed.

Nikita Marshalkin, head of machine learning at Sumsub, said modern deepfakes have become too advanced for human eyes alone to detect.

“Modern deepfakes can no longer be detected by the human eye, and decision-making should be based on multiple signal analysis in real time,” Marshalkin said.

He added that fraud prevention systems must now evolve from static verification tools into adaptive intelligence systems capable of learning continuously from new attack methods.

The development reflects a broader global shift in cybersecurity, where artificial intelligence is increasingly being used by both defenders and cybercriminals in an escalating digital arms race.

Read also: TikTok invests $200,000 in AI media literacy as Africa faces rising deepfake risks

For African businesses, experts say the challenge is becoming more urgent as digital adoption accelerates faster than cybersecurity readiness in many markets.

Fintech firms, online lenders, payment companies, crypto exchanges and gaming platforms are expected to face mounting pressure from regulators and customers to strengthen identity verification systems against AI-generated fraud.

The rise of deepfake-enabled attacks could also increase compliance costs for digital businesses already grappling with cyber risks, financial fraud and stricter anti-money laundering requirements.

With AI-generated fake videos, cloned voices and manipulated identities becoming harder to detect, companies are now being forced to invest in more advanced fraud monitoring technologies to avoid financial losses and reputational damage.