What Are Deepfakes?
Deepfakes are synthetic media created using artificial intelligence — specifically deep learning techniques — to generate realistic video, images, or audio of real people. The term combines 'deep learning' and 'fake.' Early deepfake tools required significant technical expertise and computing resources. Today, consumer-grade tools can generate convincing synthetic media from a small number of photos or a short audio clip.
Deepfakes are not inherently malicious — they have legitimate uses in film, entertainment, and accessibility. But the same technology is increasingly being used for fraud, impersonation, harassment, and disinformation.
How Deepfakes Are Used in Fraud
CEO fraud and business email compromise: Criminals have used deepfake video and audio to impersonate executives in video calls, instructing employees to transfer funds or share sensitive information. In documented cases, employees transferred significant sums after receiving what appeared to be a video call from their CEO.
Romance scams: Fraudsters use deepfake video to impersonate attractive individuals in video calls, building fake romantic relationships to eventually request money or personal information from victims.
Investment fraud: Deepfake video of celebrities or financial experts has been used to promote fraudulent investment schemes, lending false credibility to scams.
Identity verification bypass: Some deepfake tools are designed specifically to defeat video-based identity verification systems used by financial institutions and government agencies.
How to Recognize a Deepfake
Deepfake detection is an active area of research, and no detection method is fully reliable. However, some indicators may suggest synthetic media: unnatural blinking patterns or eye movement, inconsistent lighting on the face versus the background, blurring or distortion around the edges of the face or hair, unnatural skin texture, and audio that does not quite match lip movements.
These indicators are becoming less reliable as deepfake technology improves. The most reliable approach is not visual detection but behavioral verification: if someone contacts you unexpectedly and requests money, sensitive information, or urgent action, verify their identity through a separate, trusted channel — such as calling a known phone number — before responding.
Protecting Yourself from Deepfake Scams
Establish a verification protocol with family members and colleagues. Agree on a code word or a specific verification question that only the real person would know. If you receive an unexpected request — even in a video call — use this protocol before taking any action.
Be skeptical of urgency. Deepfake scams, like most fraud, often create a sense of urgency to prevent you from thinking clearly or verifying the request. Legitimate requests from banks, employers, and family members can almost always wait a few minutes for verification.
Limit the publicly available video and audio of yourself. While this is not always practical, reducing the amount of publicly accessible media makes it harder to create a convincing deepfake of you.
Sources & References
- FBI IC3 — Deepfakes and Fraud
- CISA — Contextualizing Deepfake Threats to Organizations
- FTC — Impersonation Scams
CyberIdentityTheft.com is an independent consumer education resource and is not affiliated with any of the organizations listed above.