What Is the Liar's Dividend?

What is the liar's dividend?

The liar's dividend is the strategic benefit that flows to dishonest actors when the existence of convincing AI-generated content makes genuine evidence dismissible as fake. Coined by legal scholars Bobby Chesney and Danielle Citron in 2019, it describes a second-order harm of synthetic media that often outweighs the harm of any individual fake.

Term coined in 2019 by Bobby Chesney and Danielle Citron in the California Law Review.
The first-order harm is fakes. The second-order harm is plausible deniability for real footage.
Verification is slow and expensive. Generation is fast and cheap, an asymmetry that benefits liars.
Affects journalism, courts, elections, and personal accountability.
Belief persistence means debunked claims continue to shape perception long after correction.

Origin of the term

The phrase appears in "Deep Fakes: A Looming Challenge for Privacy, Democracy, and National Security," a 2019 California Law Review article by Bobby Chesney and Danielle Citron. Chesney and Citron argued that the most damaging consequence of convincing synthetic media is not the deception of audiences with fakes but the cover it gives to liars to dismiss real evidence. Once a population accepts that any video or audio could plausibly be fabricated, every recording becomes contestable, and the burden of proof shifts away from the person caught on tape.

Why the asymmetry matters

Generating a plausible fake takes minutes and costs almost nothing. Verifying that a real recording is authentic takes specialized analysis, time, and expertise, and even confident verification rarely catches up with the speed at which doubt spreads. This asymmetry is structural: it favors the side willing to assert that any inconvenient evidence is fabricated. The result is not just more lies. It is a slow erosion of the evidentiary baseline that journalism, accountability, and democratic deliberation depend on.

Where the effect is already visible

Politicians have begun routinely dismissing real, recorded statements as AI-generated. Courts increasingly face challenges to authentic evidence on the grounds that it could plausibly be synthetic. Journalists report a rising verification burden on stories that would once have spoken for themselves. The Brookings Institution and other research groups have documented the pattern across multiple jurisdictions. Detection accuracy is also declining as generation models improve faster than detection tools.

What helps

There is no clean fix. Provenance standards (such as C2PA's content credentials), platform labeling, and improved media literacy each address part of the problem. None addresses the core asymmetry: that doubt is cheaper to manufacture than certainty is to establish. Treating recordings with appropriate skepticism, while not collapsing into total suspicion, is closer to a posture than a policy. Understanding the term is the first step toward recognizing when it is being exploited.

Where this appears in the book

Chapter 11: Generative AI and the End of Seeing Is Believing