Accounting for 22% of generative AI-driven fraud attempts, SA has the highest share of these types of incidents in southern Africa. This is according to international data, analytics and tech firm LexisNexis Risk Solutions, which notes a rising trend of generative AI (GenAI) tools used for fraud globally.
The company cites statistics from its automated identity verification solution IDVerse, which shows that in 2025 it stopped 400 000 deepfakes. Over a third of fraud incidents in the past year involved GenAI, it says – a notable increase from 20% the year before. In addition, 85% of identity fraud cases involved GenAI-created forgeries.
Jason Lane-Sellers, director of fraud and identity at LexisNexis Risk Solutions, said the company’s research showed an increase in fraudsters who manipulate documents, images and videos to impersonate people for identification purposes.
“If I can utilise AI tools to fake a form or fake a driver’s licence or trigger a view of a driver’s licence, that’s something I can use,” said Lane-Sellers.
The problem is not limited to targeted impersonation; AI is being used widely in various types of fraud and scams. A notable challenge for SA is the growth of digital channels for financial services, resulting in a rise of quick, frictionless payments and value transfers. This makes the country especially attractive to fraudsters.
“Fake endorsements, advertising, imagery, documents, invoices… [these all] build that kind of perfect storm scenario, especially if you tie it in with South Africa, where you've also got instant payments and instant money movement,” he said.
Mark Walker, director of technology consultancy T4i, agreed that SA's fintech and banking sectors have seen a sharp escalation in AI-assisted fraud over the past 18 months.
“TransUnion Africa reports a 1 200% year-on-year increase in deepfake incidents, with banking and fintech identified as the hardest-hit sector. Meanwhile, Smile ID recorded a rise from fewer than 200 deepfake attacks per month in 2024 to more than 3 000 by late 2025 across southern Africa,” said Walker.
He highlighted various examples, from AI-assisted advertising to targeted personal impersonation. One example of the use of deepfakes, he said, was part of the R2 billion penalty imposed on online trading platform Banxso by the Financial Sector Conduct Authority. While various charges were brought in that case, media reports noted that Banxso’s client acquisition model included AI-generated promotional videos featuring Elon Musk and Johann Rupert spread on social media.
For targeted use, Walker noted First National Bank’s recent alerts to customers about AI-generated voice, video and text impersonations of bank staff and family members used to authorise urgent transfers.
Walker said this trend will continue to gain momentum and evolve as new AI-based mimicking capability improves.
“As always, sustained consumer education and visible action on reported incidents are essential, as informed users remain the sector's most effective first line of defence,” Walker added.
Share
