The Financial Sector Conduct Authority (FSCA) is stepping up enforcement against artificial intelligence (AI)-powered financial scams, as deepfakes become one of SA’s fastest-growing regulatory challenges.
The regulator's newly-released Regulatory Actions Report for the 2025/26 financial year identifies online financial harm as one of its highest enforcement priorities.
It warns that increasingly sophisticated deepfake advertising, impersonation fraud and AI-generated investment scams are exposing consumers to unprecedented risks.
Rather than treating deepfake scams as isolated incidents, the FSCA says it has positioned them as a core regulatory risk that demands stronger investigations, tougher penalties and closer collaboration with law enforcement agencies.
The report shows the regulator finalised 678 investigations during the year, up from 633 in the previous financial year, while imposing administrative penalties totalling R2.89 billion on 76 individuals and entities.
The sharp increase was driven largely by serious contraventions involving financial advisory misconduct, market abuse and anti-money-laundering failures, it says.
“Deepfakes have become an increasingly effective tool for fraudsters because they allow criminals to create convincing fake video endorsements using the identities of well-known public figures, making fraudulent investment schemes appear legitimate,” says the FSCA.
The regulator says these scams are often accompanied by impersonation fraud, misleading social media advertising and unrealistic investment promises designed to lure consumers into high-risk or entirely fraudulent products.
"The increasing sophistication of scams, impersonation fraud, deepfake advertising, unauthorised forex schemes and digitally-enabled financial misconduct poses growing risks to South African consumers," the report states. “Enforcement action against these scams demonstrates the regulator's willingness to use its full regulatory powers to combat online financial harm.”
In several cases, even the FSCA, its commissioner and staff members were impersonated to lend credibility to fraudulent schemes.
The watchdog says deepfake-enabled scams require stronger collaboration between regulators, financial institutions and law enforcement.
Deepfakes go mainstream
Meanwhile, a newly-released Digital Wellness in South Africa report, conducted by the Bureau of Market Research, found that deepfakes are no longer a future cyber security concern, but an emerging threat already affecting thousands of South Africans.
The study found that 13% of its 720 respondents have been targeted by criminals using deepfake technology, placing AI-generated deception alongside financial scams, phishing and identity abuse as one of the country's growing online safety risks.
The findings come as generative AI tools make it easier and cheaper to create convincing fake content that can be used for fraud, impersonation, extortion and misinformation.
"Approximately 13% of respondents reported having experienced deepfake misuse involving manipulated images, audio, or videos. As deepfake technologies become increasingly accessible and sophisticated, this finding suggests that AI-enabled forms of digital harm are already affecting a measurable proportion of the population."
While financial scams and extortion remain the most common form of online exploitation, affecting 39.2% of respondents, the report identifies AI-generated deepfakes as one of several emerging cyber threats gaining traction in SA. Other reported incidents include phishing, identity impersonation, cyberbullying, unauthorised access to bank accounts and online grooming.
Despite the growing sophistication of AI-enabled attacks, nearly three-quarters (74.4%) of respondents said they were confident they could recognise misinformation and online scams. However, the index cautions that this confidence may provide a false sense of security.
"These findings suggest that most South Africans perceive themselves as capable of navigating digital information environments and identifying potentially deceptive content. However, confidence does not necessarily translate into actual detection ability, particularly given the increasing sophistication of misinformation campaigns, phishing schemes and AI-generated content,” the index warns.
Share
