Deepfakes and AI-Generated Identity Theft A Cybersecurity Analysis of Public Figure Impersonation on Social Media

In the past, deepfakes were a novelty on the internet, but today they have become tools of cybercriminals. Rather than treating deepfakes as a purely technological or media problem, this paper examines them as a cybersecurity threat, focusing on identity theft and the impersonation of public figures on social media. Based on a systematic review of peer-reviewed journal articles and conference proceedings from 2022 to 2026 and documented incidents, the paper discusses generation methods, deepfakes as a social engineering tool, human and automated detection, and legal responses. The results demonstrate that humans detect synthetic faces and voices at close to chance rates, automated detectors cannot accurately detect content from unseen generators, and attackers gain scalable distribution through paid advertising and fake profiles. A common five-step attack chain is shared by a corporate video-call fraud in Hong Kong, an attempted voice-clone attack on a global advertising firm, deepfake investment scams in Malaysia and an election deepfake in Slovakia. The paper's contribution is a multi-layered defence framework, which combines technical, platform, organisational, legal and user-awareness controls, alongside an agenda for future research. The paper addresses deepfakes, social engineering, identity theft, impersonation and social media security.