Faces of Fraud

With trillions of dollars stolen every year by sophisticated fraudsters, Wunderman Thompson have created a global fraud prevention campaign (for HSBC) to educate the public about the tactic’s fraudsters use to scam them.

Fraud is a growing problem, with global losses from payment fraud expected to hit £40.6 billion by 2027, a 25% increase from 2020 levels. In a world’s first, ‘The Faces of Fraud’ campaign uses a pioneering AI that can identify characteristics, such as gender, ethnicity, weight and age to predict what a fraudster’s face might look like based on voice data.

Wunderman Thompson worked together with Carnegie Mellon University, to feed in voice recordings of actual fraudsters into the ‘voice to face’ neural network to generate face composites that predict facial features based on the speaker’s unique ‘audible DNA’.

The predicted faces of the fraudsters informed the creation of various digital humans using Epic Game’s Unreal engine creation platform ‘Metahumans’. Wunderman Thompson then brought these digital ‘fraudsters’ to life using motion capture.

The ‘faces of fraud’ share their fraud tricks with the public in a series of on-line tutorials which finally put a face to a faceless crime, from romance to investment fraud.