Deepfakes and Synthetic Media

Deepfakes and AI-generated media have created a new information challenge: increasingly, you cannot trust your eyes and ears. What does this mean for society?

What deepfakes are

Deepfakes are synthetic media — video, audio, or images — in which a person appears to say or do something they never did. The technology uses AI to convincingly map one person face onto another body, or to clone someone voice from a small sample of recordings.

Why it matters now

Until recently, creating convincing fake video or audio required significant technical skill and expense. Now the barrier has dropped dramatically — free tools can generate passable results in minutes. This has shifted deepfakes from a specialist concern to a mainstream one.

Uses and misuses

Deepfakes are used for entertainment (dubbing films into different languages), satire, and legitimate creative work. They are also used for non-consensual intimate images (a serious harm, and increasingly illegal), political disinformation, fraud, and harassment.

The epistemic problem

The most significant societal consequence may not be specific deepfakes, but a general erosion of trust in media. When video and audio can be fabricated, people may start dismissing genuine evidence as fake — a phenomenon researchers call the "liar dividend".

Detection and authentication

Researchers are developing AI detection tools and content authentication standards (like the C2PA standard, which embeds provenance information in media files). These are imperfect and still catching up to generation capabilities.

Synthetic media forces a fundamental question: what does it mean to have evidence of something? The answer will reshape law, journalism, and public discourse over the next decade.