Fraud is on the rise in South Africa, with much of it relating to falsified documents and fake credentials. Whether the fraudsters’ aim is to secure goods on credit, or inflate insurance payouts, the end result is serious financial and reputational losses for South African businesses.
The Association for Savings and Investment South Africa reported that in 2023, South African life insurers and investment companies detected 13 074 cases of fraud and dishonesty, a 46% increase from the previous year, costing the industry at least R175.9 million.
The South African Banking Risk Information Centre Annual Crime Statistics Report for 2023 revealed a loss of almost R3.3 billion, including a 100% increase in vehicle asset finance fraud.
Moonstone Information Refinery cites a global Sumsub report stating that identity fraud involving forged passports and identity documents accounted for 50% of all fraud attempts worldwide in 2024, with South Africa emerging as one of the hardest-hit countries.
In a bid to mitigate these risks, organisations have long resorted to rigorous know your customer (KYC) background checks − generally carried out by teams of in-house employees or outsourced partners. However, manual checks can take time, which negatively impacts customer experience.
As fast as criminals have moved to harness AI to commit fraud, vendors are training AI models to detect document fraud.
For example, if a person is applying for a job, the prospective employer would have to call the university to check whether the degree is legitimate and the marks on the document are accurate. It’s a long, tedious process.
Bank statements are another example. People have to supply bank statements as part of any credit check process. Say your income is R35 000, it’s fairly easy to make the 3 look like an 8 to increase your income to R85 000 to boost creditworthiness and run up massive bills you cannot afford to pay back.
AI makes deception harder to detect
With the advent of AI, manual checks are at risk of being less accurate, since AI is being used to alter documents in ways that are increasingly difficult for humans to detect.
Whether the documents are invoices, IDs, proof of residence, bank statements, death certificates or university degrees, it has become even easier for fraudsters to generate documents that the human eye cannot detect as fraud.
One example of this was OnlyFake, a website which allowed anyone to quickly and easily generate very convincing ID documents, passports and driver’s licences, and even batch-create hundreds of them at a time.
However, as fast as criminals have moved to harness AI to commit fraud, vendors are training AI models to detect document fraud. AI-enabled verification solutions are now capable of verifying documents using hundreds of checks, in a matter of minutes.
The technology assesses metadata, internal structures, image inconsistencies, fonts and reused, generated or/and ‘template farmed’ documents, to produce a risk rating for each document.
Automation alongside AI can identify low-quality images which can be a cover for document forgery and manipulation, and automatically request customers to send a better-quality version of the document.
Whether this AI is built into a platform or offered as-a-service, it quickly confirms changes or highlights suspicious elements for further investigation. This enables faster, more reliable customer and merchant onboarding, credit applications, employee screening and insurance claims management.
Capable of up to 99% accuracy in document verdicts and reducing KYC onboarding time by up to 60%, AI-enabled document forensics automates fraud detection and catches fraud attempts in seconds, with risk capabilities to flag the level of risk associated with anomalies.
With AI and automation, the organisation is able to run far more thorough checks to better manage risk, and is able to do so more cost-effectively than when using manual processes.
Win-win-wins for better business
With automated, AI-enabled document forensics and fraud checking, organisations are able to reduce fraud and financial losses, improve efficiencies and enhance the customer experience. Customers no longer have to wait days or weeks for their credit application to be approved.
The new way aligns with the digitisation and automation journeys that companies around the world are on. Customers today don’t want to walk in with a folder full of their KYC documents − they want to quickly upload them onto portals and get rapid responses.
It has become important to meet customer expectations while also managing risk, and AI and automation allow organisations to do so.
At the same time, they offer the benefits of lower overheads and reduced fraud losses, with improved accuracy and the ability to improve financial forecasting.
Share