For years, a video call has been one of the easiest ways to build trust remotely. You see someone, they see you, and a decision gets made. Identity verification firm Identomat argues that this assumption is now a liability. Generative AI lets fraudsters convincingly impersonate executives, customers and business partners using publicly available footage and voice recordings.

The danger is no longer hypothetical. In early 2024, an employee at a multinational firm in Hong Kong joined what appeared to be a confidential video conference with the company’s UK-based chief financial officer and several finance colleagues. None of them were real. According to Hong Kong police, the call was built with deepfake technology from public video and audio clips. The employee authorised transfers of roughly HK$200m, about $25.5m, to five local bank accounts. The company was later named as engineering group Arup, which confirmed that fake voices and images had been used.

The FBI has issued similar warnings about criminals using AI-generated video and audio to impersonate executives and other trusted individuals. As these attacks become increasingly sophisticated, organisations need to look beyond what they see and hear on screen. Video can be a valuable part of the verification process, but it should never be the sole proof of identity.

Automated KYC has transformed digital onboarding, letting customers submit documents, take selfies and pass liveness checks without speaking to anyone. That works for most straightforward cases. The gaps appear elsewhere: inconsistent customer data, enhanced due diligence, unusually large transactions, account recovery from new devices, and internal requests such as an executive asking to move millions or change beneficiary details.

Source link