Explore our new Cybersecurity Terms Glossary, explained by our experts.

View the Glossary View the Glossary
PRIVIA

Cybersecurity Terms Glossary

What Is a Deepfake?

A deepfake is synthetic audio, video, or imagery generated with deep learning to convincingly imitate a real person, now widely used in fraud, executive impersonation, and disinformation.

Illustration for the cybersecurity term Deepfake
Home / Blog / Cybersecurity Terms / What Is a Deepfake?
17 November 2026

What Is a Deepfake?

What Is a Deepfake?

A deepfake is audio, video, or imagery generated or manipulated with deep learning so that it convincingly shows a real person saying or doing something they never did. The term comes from a Reddit user who posted face-swapped videos in 2017, but the technology has moved far beyond novelty: voice clones and real-time video avatars are now practical tools for fraud, and they target organisations as much as individuals.

How Deepfakes Are Made

Early deepfakes relied on autoencoders and generative adversarial networks (GANs), in which one neural network generates fake content and another tries to tell it apart from the real thing. Each round of that contest makes the output more convincing. Diffusion models and large speech models have since pushed quality higher while lowering the amount of training data required.

Voice is the most worrying area for businesses. Several commercial and open-source tools can now clone a voice from a short clean recording, and senior executives usually leave plenty of material online in earnings calls, conference talks, podcasts, and social media videos. Real-time face-swap software can overlay another person’s face onto a live webcam feed, making a video call no longer proof of who is on the other end.

How Attackers Use Them

Executive Impersonation and Payment Fraud

Deepfakes give business email compromise a new layer of credibility. In 2019, the chief executive of a UK-based energy firm transferred €220,000 after a phone call in which criminals used AI to mimic the voice of his parent company’s boss, according to the firm’s insurer. In early 2024, an employee in the Hong Kong office of the engineering firm Arup paid out around HK$200 million after a video conference in which the chief financial officer and other colleagues were all deepfakes. In May 2024, criminals set up a fake meeting using a voice clone of WPP’s chief executive in an attempt to defraud staff; it failed because employees were suspicious.

Bypassing Identity Verification

Remote onboarding and account recovery often rely on a selfie video or a voice check. Attackers use face-swaps, replayed footage, and virtual camera software to inject synthetic video into these flows. Banks and identity providers now test their liveness detection against such injection attacks, guided by standards such as ISO/IEC 30107 for presentation attack detection.

Fake Candidates and Insiders

Security agencies have warned about job applicants using deepfaked video in remote interviews to win IT roles that give them access to company systems, a tactic linked to North Korean IT worker schemes.

Reputation and Disinformation

Fabricated clips of executives or politicians can move share prices, damage brands, or influence public opinion before any correction catches up.

Why Detection Alone Is Not Enough

Visual clues such as unnatural blinking, mismatched lighting, or blurred edges around the face were useful a few years ago, but modern models have largely eliminated them. Automated detectors exist, yet they struggle with compressed video, new generation methods, and real-time calls. Provenance standards such as C2PA content credentials, which attach a signed history to media, help prove that genuine content is authentic, but they cannot label fakes that were never signed. The EU AI Act requires deepfakes to be disclosed as artificially generated, and in the UK the Online Safety Act 2023 made sharing intimate deepfake images without consent a criminal offence—but criminals committing fraud will not label their work.

Controls That Actually Work

The most reliable defence is a process that does not depend on recognising a face or a voice:

Deepfakes are ultimately a social engineering tool, so the best way to measure your exposure is to test it. Our social engineering services include voice-based impersonation scenarios that show whether your payment and helpdesk procedures hold up under pressure.

FAQ – Frequently Asked Questions

How Much Audio Does Someone Need to Clone My Voice?

Far less than most people expect. Current voice cloning tools can produce a convincing imitation from a short, clear recording, and quality improves with more material. Executives who speak at conferences, appear in videos, or host podcasts provide ample training data. Because limiting this material is rarely realistic, organisations should focus on verification processes that do not rely on recognising a voice.

Can I Spot a Deepfake on a Video Call?

Sometimes, but you should not rely on it. Asking the person to turn their head fully sideways, pass a hand in front of their face, or answer a question only the real person would know can expose weaker fakes. Modern real-time tools are improving quickly, though, so treat any unusual or urgent request made on a call as unverified until confirmed through a separate channel.

Are Deepfakes Illegal in the UK?

Creating a deepfake is not automatically illegal, but using one usually is. Deepfake-enabled fraud falls under the Fraud Act 2006, and impersonation for financial gain is prosecuted like any other fraud. The Online Safety Act 2023 made sharing intimate deepfake images without consent an offence. Defamation and data protection law may also apply when a real person’s likeness is misused.

Related Terms

You May Be Interested In These

What Is Wireless Penetration Testing?

Cybersecurity Terms

17 November 2026

What Is Wireless Penetration Testing?

Read More
What Is Mobile Application Management?

Cybersecurity Terms

17 November 2026

What Is Mobile Application Management?

Read More
What Is S3 Object Lock and Immutable Backup?

Cybersecurity Terms

17 November 2026

What Is S3 Object Lock and Immutable Backup?

Read More