TechRadar analysis questions whether FCA is underestimating AI fraud

TechRadar’s Bharat Mistry questions whether the Financial Conduct Authority and financial institutions are paying enough attention to the speed of AI-enabled fraud.

The analysis says generative AI can make convincing voices, faces, identity documents and supporting paperwork cheaper and easier to produce. Existing controls such as video liveness checks, voice callbacks and one-off document verification are not described as useless, but their effectiveness can no longer be assumed.

Mistry gives particular attention to synthetic identity fraud, in which genuine information is combined with invented details to create an apparently legitimate individual. AI can help build the documentation and digital footprint needed to make that identity credible. Because the person may not exist, there may be nobody to report the fraud, and a synthetic customer can establish a financial history before committing fraud later.

The article argues that banks should test their own identity and onboarding systems with the same adversarial approach used for networks and applications. It points to questions about AI-generated voices, synthetic faces, fabricated documents and identities assembled across multiple data points. It also says behavior over time may provide stronger signals than one-off verification. The FCA has an important role, but the analysis says regulation alone cannot anticipate every emerging technique. It calls for continuous testing, senior accountability and explainable automated decisions, warning that AI governance should not become paperwork that leaves institutions exposed.