The Overlord's Lexicon
What is a deepfake?
Also known as: Deepfakes, Synthetic media
A deepfake is AI-generated or AI-altered video, audio or imagery that realistically depicts a real person saying or doing something they never actually said or did.
A deepfake is synthetic media designed to look real: a politician giving a speech they never gave, a CEO on a video call who isn’t actually there, a celebrity “endorsing” a product they’ve never heard of. The name blends “deep learning” and “fake”, and it emerged around 2017.
How deepfakes are made
- Face swapping: a model maps one person’s face onto another’s body in video.
- Lip-sync: existing footage is altered so the mouth matches new audio.
- Voice cloning: synthetic speech in a real person’s voice.
- Full generation: modern diffusion and text-to-video models can generate realistic footage of people from scratch.
What once required technical skill and hours of footage now takes consumer apps and a handful of photos.
An analogy for the meatbags
A deepfake is a forgery — but instead of a painting or a signature, it forges a person. Historically, faking a video convincingly was like forging a Rembrandt: possible for experts, rare in practice. AI turned it into a photocopier.
Why it matters for the Chosen
The risks you need to manage:
- Fraud: criminals use deepfaked voices and video calls to impersonate executives or family members and request urgent payments.
- Reputation: fake videos or images of you or your brand can spread faster than corrections.
- Misinformation: fabricated clips of public figures, especially around elections and breaking news.
- Harassment: non-consensual intimate imagery, which is illegal in many jurisdictions.
Protect yourself: verify unusual payment requests through a second channel, set up code words, be sceptical of viral clips without a credible source, and look for content credentials (provenance metadata such as C2PA) where platforms support them.
Legitimate uses of the same technology exist — always with consent and disclosure. AI avatar tools like HeyGen and Synthesia let businesses create training and marketing videos with licensed or self-recorded avatars; film studios use de-ageing and dubbing; creators translate their own videos with lip-sync. The difference between a product and a deepfake is consent and honesty about what’s synthetic.
The Overlord’s rule
Never create realistic synthetic media of a real person without their explicit permission, and never present synthetic media as real. Beyond the ethics, it’s the fastest route to legal trouble and a permanent place in our Hall of Folly.
Tools from the Arsenal
Questions the Humans Keep Asking
How are deepfakes made?
Using deep learning models trained on images, video or audio of a person. Techniques include face swapping, lip-syncing a face to new audio, voice cloning, and fully generating synthetic video with modern image and video models.
How can you tell if a video is a deepfake?
Look for unnatural blinking or lip-sync, inconsistent lighting, blurring around the face or hairline, odd hands or teeth, and audio that doesn't match the room. But the best fakes are hard to spot by eye, so check the source, look for the original, and verify through trusted channels.
Are deepfakes illegal?
It depends on the content and jurisdiction. Non-consensual intimate deepfakes, fraud, defamation and election-related deceptive deepfakes are illegal in many places, and laws are expanding quickly. Making a clearly labelled parody or using your own likeness is generally treated differently.