You saw it with your own eyes. Your eyes were shown a rendering.
Also known as Deepfakes
Audio, images, or video generated or altered by machine learning to depict things that never happened: a voice clone of a relative asking for money, a video of a public figure saying words they never said, a photograph of an event that never occurred. The danger is twofold. The obvious half is the forgery itself, which exploits the deep human habit of trusting what we see and hear. The quieter half is what forgeries do to real evidence: once anyone can claim a damaging recording is 'just a deepfake,' the existence of the technology becomes an escape hatch for the guilty. The same tool that fabricates the false also laundered the true.
Truth-adjacency
Truth-independent: the pattern works regardless of whether the claim is true
Where it shows up
Platforms and algorithms
How it works
The phrases and tells that mark this pattern in the wild:
a recording that arrives with no verifiable sourcea voice that sounds almost right but slightly offa video whose lighting, blinking, or lip-sync misbehaves on close watcha claim that a real recording is AI-generated, offered with no evidencemedia that appears first on anonymous channels and spreads before it can be checkedSynthetic media attacks the two habits you rely on most: trusting your senses and trusting your sources. The forgery side shows up as media that is emotionally urgent, source-less, and resistant to verification, often timed to do maximum damage before it can be checked. The dividend side shows up as a reflexive ‘that’s AI-generated’ offered by someone with a stake in the media being false, with no technical analysis behind it. In both cases the tell is the same: the conversation is being steered away from the chain of custody. Where did this come from, who released it, and can anyone independent confirm it? When those questions are treated as obstacles rather than the point, the pattern is at work.
You call “synthetic media” on a genuine recording because it is damaging to someone you support, offering the deepfake label as a blanket denial rather than a demonstrated fact. Real evidence can be unflattering and still real. The pattern requires either fabrication or the strategic use of the fabrication claim. If the media survives source verification and independent corroboration, it is real, and dismissing it is denial, not detection. If the fabrication claim is backed by actual analysis of the media’s artifacts, that is legitimate scrutiny, not the dividend.
One of these two real scenarios is Synthetic media. The other is a different pattern entirely. Which one is which?
The tell
A recording or image that is verified through its source, its metadata, and independent corroboration is real evidence, however imperfect. Not everything that looks strange is synthetic, and not everything that looks clean is real. The pattern requires the media to be machine-fabricated or the claim of fabrication to be used as a shield. If the media checks out through its chain of custody, it is evidence. If the 'it's a deepfake' claim comes with no analysis and conveniently serves the person it would exonerate, that is the dividend, not a diagnosis.
Convincing fabrication becomes available to anyone. Every recording, real or not, now sits under a doubt that used to require specific evidence to raise.
Because the habit of trusting what you see is older than any technology and cannot be switched off by knowing better. Suspending it entirely costs you the real recordings too, which is the outcome the forgers and the deniers both benefit from.
The forgeries did less damage than the doubt they made available, and the doubt was free to anyone who needed it.
Field notes where this pattern was identified:
How this pattern gets misused
Someone dismisses any inconvenient recording or image as a deepfake without evidence, treating the mere existence of the technology as proof that any specific piece of media is fake. This is the liar's dividend wearing a lab coat. The term becomes a universal solvent for accountability, where the burden of proof is flipped and real evidence is discarded on speculation.
What it looks like when you're wrong about it
A recording or image that is verified through its source, its metadata, and independent corroboration is real evidence, however imperfect. Not everything that looks strange is synthetic, and not everything that looks clean is real. The pattern requires the media to be machine-fabricated or the claim of fabrication to be used as a shield. If the media checks out through its chain of custody, it is evidence. If the 'it's a deepfake' claim comes with no analysis and conveniently serves the person it would exonerate, that is the dividend, not a diagnosis.
Not sure? Describe the situation to someone outside it. If they do not see the pattern, pause before you name it.
Synthetic sincerity
It sounded like it meant it.
Authority laundering
You stopped asking for the evidence because the phrase sounded settled. 'Experts agree' was the whole argument. The authority was borrowed, never shown.
Liar's dividend
The existence of fakes lets anyone dismiss real evidence as fake.
Reality apathy
You did not decide the story was false. You decided that finding out was not worth the afternoon, and moved on.
The name is designed to spread. The hook is designed to stick. If you recognized something, share the name.
Seen a real example of synthetic media? Suggest it for the Register →