CDD and Generative AI – Risks of KYC Fraud

Generative artificial intelligence (GenAI) is revolutionising industries by unlocking efficiencies, redefining what is possible, and creating highly realistic text, images, and voices. While these advancements offer immense benefits, they also present new challenges to customer due diligence (CDD) and fraud prevention processes. Fraudsters are exploiting GenAI to bypass security measures implemented by financial institutions (FIs), making it imperative for CDD functions to innovate and adapt rapidly.

The financial sector is experiencing escalating fraud cases amid the digitisation of payments and onboarding channels, notably influenced by the adoption of Faster Payment Systems. In 2022, banks incurred losses of £1.2 billion to payment fraud. Despite institutions intercepting 61.5p of every £1 of fraud, the emergence of GenAI deepfakes has given rise to a new wave of impersonation fraud that existing systems are ill-equipped to handle.

Within the context of fraud, deepfake technology refers to the sophisticated use of GenAI tools to create or manipulate audio and visual content to fabricate identity and evidence. This technology enables fraudsters to convincingly impersonate legitimate individuals or create realistic but fraudulent documents to bypass traditional security measures and perpetrate financial or identity fraud.

To counter the evolving threats posed by GenAI, FIs must strengthen their control frameworks by enhancing existing measures and integrating advanced combative technologies within their CDD processes. To manage the risks associated with deepfakes, FIs can deploy a multidisciplinary approach, leveraging a combination of available technologies. For instance:

• Subject Matter Expertise on both CDD and AI can identify hotspots of KYC Fraud within the onboarding process, design Target Operating Model that incorporates safeguards against deepfakes and falsified documents, offer training to staff on the latest regulatory technologies etc
• Advanced biometrics, including facial recognition, voice identification, and behavioural biometrics, based on measurable patterns of human activities, form a safer authentication toolkit that is more difficult to be circumvented with deepfakes
• Forward-looking strategy can explore the application of AI as a countermeasure to analyse metadata and internal integrity for signs of forgery, blockchain as a decentralised and tamper-proof ledger etc

With the Payment System Regulator’s Faster Payments reimbursement requirements taking effect in October 2024, compensation will be shared between the paying and receiving Fis, with each liable for half of the victims’ losses. If they fail to prevent the onboarding of fraudulent or mule accounts used in the laundering of funds defrauded through advanced push payment fraud, they will bear evermore significant financial consequences.

Therefore, investment in expertise and technology to futureproof KYC should not be an afterthought, but an imminent necessity. Ultimately, there will be a need to fundamentally address vulnerabilities within CDD process and integrate these mitigation controls.

KPMG assists organisations in designing customised solutions to combat the threats posed by GenAI, providing Subject Matter Expertise in CDD, Fraud, and AI Strategy. By integrating resources from these domains and adopting a multidisciplinary approach, KPMG leverages its global AI Community to offer AI-driven detection algorithms, enhanced risk assessment methodologies, and innovative fraud prevention strategies. Current offerings include CDD Advisory, financial crime training focused on KYC and Fraud, a Target Operating Model for Fraud, and KPMG’s AI Strategic Framework. Stay tuned for our upcoming series on CDD and GenAI.