Liar's Dividend

What the liar’s dividend is

The liar’s dividend is the advantage a dishonest person gains once the public knows that realistic fakes are possible: when caught on a genuine recording, they can simply wave it away as a deepfake. The term was coined by the legal scholars Bobby Chesney and Danielle Citron in a 2019 paper for the California Law Review, and it names a danger most people miss when they think about synthetic media.

The obvious worry about fakes is that they will fool us into believing things that never happened. The liar’s dividend runs the other way. Ordinary misinformation creates false content to make you believe a lie. The liar’s dividend destroys trust in true content so that you disbelieve the truth. It does not require the liar to fake anything at all. They only need to invoke the possibility of fakery, and let the doubt do the rest.

The practical effect is a flip in the burden of proof. For a long time, a clear recording shifted the weight of an argument onto the person it caught: the footage existed, so they had to explain it. The liar’s dividend hands that weight back. Now the person holding genuine evidence has to prove it is real, which is a far harder thing to do than to gesture vaguely at the chance that it is not.

Turning doubt into an alibi

What makes the dividend so useful is that it is paid out by the general atmosphere, not by any single fake. Once everyone has heard that convincing fakes exist, “that is fake” becomes a move anyone can make about anything, at no cost and with no evidence. The liar does not need a forgery in hand. They need only the public’s awareness that forgeries are out there.

The shape of the move is always roughly the same. Someone in a position of trust is caught on a recording saying or doing something they would rather they had not. Instead of denying the substance, they attack the artefact: the audio is AI, the video is doctored, the whole thing is a smear built with clever software. The same trick works on ordinary people, where genuine footage of an incident gets dismissed as generated. In each case the goal is not to win the argument but to dissolve it, so that no shared fact survives for anyone to be held to.

The damage is not confined to politics. In courtrooms and everyday disputes alike, authentic photographs, recordings and documents can now be met with the same shrug, forcing the side that holds real proof to spend its energy establishing that the proof is genuine before the substance can even be discussed. Doubt that costs the liar nothing to summon can cost the honest party a great deal to answer.

Why warning people can backfire

Here is the part that makes the liar’s dividend more than a footnote. Chesney and Citron noticed something uncomfortable: the more the public learns that deepfakes exist, the stronger the dividend grows. Every honest effort to teach people “be careful, videos can be faked now” also teaches them “so this video might be faked too” - which is precisely the doubt the liar wants to borrow. “You know as well as I do that anything can be faked these days” is a near-perfect dismissal, and rising awareness of fakery is what makes it sound reasonable.

This is not just theory. A large 2024 study published in the American Political Science Review ran five experiments with more than fifteen thousand people and found that falsely crying fake about a real scandal could indeed help a public figure hold on to support. It means that media literacy taught carelessly, as a flat instruction to distrust all recordings, can hand the dividend straight to the people it was meant to protect us from.

The lesson is not to stop teaching people about fakes, but to teach the right one. “Distrust every recording” pays the dividend. “Ask where a recording came from, and who stands behind it” does not, because it gives people somewhere to put their doubt other than a blanket shrug.

Why we fall for it

The dividend needs a willing audience, and we often volunteer. When a recording damages someone we support, “it is probably fake” is the comfortable exit. This is motivated reasoning at work: we reach hardest for the explanation that spares us a conclusion we did not want. Alongside it sits confirmation bias, the habit of waving through evidence that fits our existing view while interrogating anything that threatens it. Inconvenient footage gets the full cross-examination; convenient footage gets a free pass.

So the liar supplies only half of the dividend. They offer the alibi. Our own wishes supply the rest, by making us glad of a reason to look away. That is why the dividend is collected most easily inside a group that already wanted to believe its own side was being framed. The more divided the audience, the easier it is to collect, because each side arrives already primed to read any attack on its own people as a fabrication.

Doubt as an old product

The liar’s dividend is a new delivery system for an old tactic. You do not have to win an argument if you can manufacture enough doubt that no argument can ever be settled. The tobacco industry understood this decades ago, when an internal memo reportedly described doubt itself as the product they were selling - enough uncertainty about the science, and the public would never quite act. The same fog later settled over other inconvenient findings.

Online, the firehose of falsehood works on the same principle: flood the space with so many competing claims that sorting truth from falsehood feels hopeless, and people give up trying. Add the steady drizzle of AI slop thickening that fog, and the liar’s dividend becomes the natural personal escape hatch within it. Where the firehose makes the whole environment feel untrustworthy, the dividend lets one person step through that distrust and walk away from a specific, awkward fact.

The defence: provenance, not blanket distrust

The tempting response to all this is to trust nothing - to treat every recording as suspect until proven otherwise. That instinct feels safe, but it is exactly what pays the dividend, because it makes “it’s fake” a universal and unanswerable defence. Blanket distrust does not protect you from liars. It works for them.

The better move is the same one that helps against deepfakes, only from the other side of the problem. Judge a recording by its provenance rather than your gut feeling about whether it looks real: who first published it, whether they have a record worth trusting, and whether independent sources point the same way. The aim is to keep healthy scepticism from curdling into corrosive cynicism, the state in which a person decides that everything is fake and nothing can be known. That far end of the spectrum is the mood that feeds dead internet theory, and it is just as much a failure of judgement as believing every fake you are shown. The narrow path is to trust in proportion to where something came from, and to refuse to let “it could be fake” quietly become “so nothing is true”.