For years, we have relied on familiar signals to establish trust online. A recognisable face on a video call, a familiar voice on the phone, or an email from a known colleague could give us confidence that the person on the other side was who they claimed to be. AI is changing that. It can generate faces, clone voices, and replicate writing styles, making it increasingly difficult to distinguish between genuine and fabricated interactions.
For businesses, this creates a challenge that extends beyond traditional cybersecurity. The question is how they can establish that someone is legitimate when criminals can replicate the signals traditionally used to verify identity.
The threat is already here
This is not a distant concern for South African businesses. The Annual Banking Crime Statistics Report from, found that reported losses from digital banking crime increased to R2.4 billion in 2025, up from approximately R1.9 billion in 2024. Crime related to banking apps accounted for more than 70% of reported digital banking losses. Card fraud losses also increased by 18%, reaching R1.7 billion. These figures show how financial crime is increasingly concentrated in digital environments where identity, access, and trust intersect. SABRIC’s report also highlights AI-driven fraud, cybercrime, and digital banking scams among the evolving threats facing the banking ecosystem.
AI does not need to create an entirely new form of fraud to be effective. It can make existing fraud more convincing by combining publicly available information, stolen personal details, and synthetic content.
Identity is becoming harder to verify
A convincing identity does not have to be completely fictional. Fraudsters can combine genuine information with fabricated details, stolen credentials, manipulated documents, and synthetic media to create an identity that appears legitimate across several interactions. TransUnion’s H1 Update: Top Fraud Trends found that South Africa recorded a 3% suspected digital fraud rate for consumer transactions in 2025, the highest among the African countries analysed. The rate declined from 4.3% in 2024, but TransUnion notes that the lower headline rate does not necessarily mean reduced criminal activity, as fraud has become more targeted and identity-driven.
In South Africa, suspected fraud was highest at the account-login stage, at 3%, compared with 2.4% at account creation and 0.7% for financial transactions. TransUnion describes this as a pattern unique in the region, pointing to the growing importance of protecting established accounts and identities, rather than focusing only on new account creation.
The report also found that 33% of South African consumers who reported losing money to digital fraud said the losses came from third-party seller scams on legitimate e-commerce platforms. This shows how fraud can exploit established trust within platforms and interactions that appear legitimate.
For businesses, this means a familiar face, voice, or email address should be treated as one indicator of identity rather than definitive proof. The more consequential the action, the more independent evidence should be required for verification.
Trust needs to become part of the process
The risk extends beyond fraudulent interactions. As employees become more aware of deepfakes and AI-enabled impersonation, they may also begin questioning genuine interactions. Employees could dismiss a legitimate executive making an unusual request because they believe the interaction could be fabricated. The solution is simple: verifiable trust. When an interaction carries higher risk, employees should have a clear way to confirm that the person, request, and context are genuine through a separate, trusted channel. This does not mean slowing down every digital interaction. It means applying greater scrutiny where the consequences are higher.
INTERPOL’s African Cyberthreat Assessment Report found that AI is enabling 55% of reported cybercrimes across Africa, making attacks “faster, more scalable, and increasingly difficult for victims and platforms to detect”. The report also highlights AI-facilitated scams, credential harvesting, automated social engineering, and sophisticated business email compromise.
The future of trust
AI is changing the economics of impersonation. What once required significant time, technical skill, and resources can increasingly be produced quickly and at scale. For businesses, this challenges the assumption that something familiar is probably genuine. Identity, therefore, needs to be established through more than what people see or hear. By combining authentication, behavioural, and device signals with clear verification procedures, businesses can make trust less dependent on appearance alone. AI is not making identity less important. It is raising the standard for how businesses need to verify identity.
