UCL Department of Science, Technology, Engineering and Public Policy


Dr Peter Novitzky receives funding to develop AI device to prevent call fraud

19 January 2022

The conceptual home edge IoT device, called WatchDog, aims to address the issue of voice-based social engineering attacks.

Person in white lab coat using tablet with IoT icons

Through their project RED-AID (REspectful and capability-centreD AI Device for Preventing Call Fraud), Professor Max Van Kleek, Oxford University and Co-Investigator Dr Peter Novitzky, UCL STEaPP, will utilise AI technology to help individuals with impaired autonomy, for example, a person with dementia, who can be especially vulnerable to voice-based social engineering attacks.

Voice-based social engineering attacks, where attackers use voice-based conversations for malicious intent, are increasing in frequency and the harms caused by these attacks can be devastating, both financially and emotionally. Older adults with mild cognitive impairment (MCI) remain one of the top, and largely defenceless, targets for such attacks. As AI systems are being applied by adversaries for automated open-source intelligence (OSINT) and execution, it is likely that these attacks will become even more effective.

To mitigate these harms, the team have developed WatchDog, a conceptual home edge IoT device. WatchDog monitors, transcribes, and analyses real-time landline phone conversations and identifies attacks by looking for clues in the dialogue, such as coercive language, to provide a means of immediate mitigation. The device also provides retrospective memory support, serving as a private diary of conversations to help individuals review, together with family members/carers, key details of whom they spoke to and what they disclosed.

RED-AID Project Lead Professor Max Van Kleek and Dr Peter Novitzky said: "Phone fraud (aka ‘vishing’) accounts for one of the most significant types of crime against individuals in the UK, and has become a huge source of personal financial loss. Yet, people have few defences against this form of crime, which are becoming even more effective through AI-powered methods and techniques.

"With the support of Howz, a company with an extensive portfolio of IoT devices for assistive ambient living in the home, our multi-disciplinary research team will involve key stakeholders in its design and development to ensure the device is not only effective, but respects individuals' autonomy, privacy, and dignity at all stages. Our collaboration across the socio-technical landscape aims to not only contribute to the responsible development of a novel AI-based technology addressing a pressing societal challenge but also to contribute to the formulation of a more widely applicable ethical design framework for ambient assistive devices in the home."

The project has been funded through the PETRAS National Centre of Excellence for IoT Systems Cybersecurity Internal Strategic Projects and Engagement Fund (ISPEF). PETRAS’s work considers issues of Privacy, Ethics, Trust, Reliability, Acceptability, Trust and Security as they apply to IoT systems and networks, as well as associated technologies at the edge, such as AI and Machine Learning. The ISPEF funds cutting-edge research projects look that tackle social and technical challenges of emerging technologies by transforming academic knowledge into practice. The projects provide technology solutions for the betterment of society and the UK economy, as well as identify potential issues and risks across sectors, ultimately helping to ensure the UK is a safer and more prosperous place.