DigitalMANIPULATIVE

Deepfakes/Synthetic Media

What it is

AI-generated fake audio, video, and images that are increasingly indistinguishable from authentic content, enabling unprecedented deception.

How it works

Machine learning models can now generate realistic human faces, clone voices from short audio samples, and create video of people saying things they never said. As the technology improves and becomes accessible, the potential for political manipulation, fraud, and reputation destruction grows exponentially.

Real-world examples

•AI-generated voice calls impersonating CEOs to authorize fraudulent wire transfers.
•Political deepfakes showing candidates making statements they never made, going viral before debunking.
•Non-consensual synthetic intimate imagery used for blackmail and harassment.

Historical case studies

The $25 million deepfake video conference

2024Corporate Fraud

Hong Kong police reported in February 2024 that a finance employee at a multinational company, later identified as the engineering firm Arup, had transferred about $25 million after a video call in which the chief financial officer and several colleagues all appeared to be present. Every other participant was a synthetic recreation made from publicly available video and audio. The employee had been suspicious of the initial email request for a secret transaction; the group call, with familiar faces and voices, was what removed the doubt.

Source →

The cloned-voice Biden robocall

2024US Primary Election

Two days before the January 2024 New Hampshire presidential primary, thousands of voters received a robocall in which an AI-generated imitation of President Biden's voice told them to stay home and "save your vote" for November. The calls were commissioned by Steve Kramer, a political consultant then working for a rival Democratic candidate, whose campaign said it had no knowledge of them; Kramer said he wanted to draw attention to the dangers of the technology. The Federal Communications Commission fined him $6 million for caller-ID spoofing and ruled that AI-generated voices in robocalls fall under existing law. A New Hampshire jury later acquitted him of state criminal charges.

Source →

Ethical guidelines

●Creating synthetic media of real people without consent is a serious ethical violation.
●Deepfakes used for deception, fraud, or harassment should be criminal offenses.
●All synthetic media should be clearly labeled as AI-generated.

How to defend against it

►Verify extraordinary video or audio through multiple independent sources before believing it.
►Look for deepfake artifacts: unnatural blinking, inconsistent lighting, audio that doesn't match lip movement.
►Default to skepticism for any dramatic audio/video that surfaces without clear provenance.

From the Defense Playbook

Out-of-Band Verificationminutes

Confirm any request that arrives through one channel (email, text, chat, a phone call) by reaching the supposed sender through a different channel you already trusted before the request existed.

Every playbook entry states how strong its evidence is and when not to use it. Browse the full playbook.

DISARM framework

How analysts would file this in the open DISARM framework (v1.6.1, CC BY-SA 4.0). Closest matches, mapped by Persuasion Lab.

T0086.002Develop AI-Generated Images (Deepfakes)T0087.001Develop AI-Generated Videos (Deepfakes)T0088.001Develop AI-Generated Audio (Deepfakes)

References

Chesney, R., & Citron, D. K. (2019). Deep fakes: A looming challenge for privacy, democracy, and national security. California Law Review, 107(6), 1753–1819
Foundational legal analysis of deepfake harms — fraud, non-consensual intimate imagery, election manipulation — and of the 'liar's dividend' by which real evidence can be dismissed as fake.
Vaccari, C., & Chadwick, A. (2020). Deepfakes and disinformation: Exploring the impact of synthetic political video on deception, uncertainty, and trust in news. Social Media + Society, 6(1)
Experiment with a political deepfake: viewers were not reliably deceived but became more uncertain, and uncertainty reduced trust in news — a corrective to the entry's emphasis on direct deception.
Europol Innovation Lab (2022). Facing Reality? Law Enforcement and the Challenge of Deepfakes. Europol observatory report
Law-enforcement documentation of deepfake use in CEO-impersonation fraud, extortion with synthetic intimate imagery, and disinformation — the entry's three examples.

Last reviewed 2026-09-12

Suggest a correction

Detect Deepfakes/Synthetic Media in any text

Paste any message, email, or article into our free Manipulation Detector to see if Deepfakes/Synthetic Media or other techniques are being used on you.

Try the Detector Free Create Free Account

Dark Patterns in UX: How Apps Manipulate Your Behavior

Subscription traps, misleading interfaces, and engineered addiction. Understanding the persuasion techniques built into the apps you use every day.

7 min read

OSINT for Beginners: Open Source Intelligence Explained

Open Source Intelligence (OSINT) uses publicly available data to gather actionable insights. Here is a beginner-friendly guide to what OSINT is and how it is used.

10 min read