You could often tell when a story had been passed through too many mouths—when details had been added for heat, or when certainty was being performed. The telling itself carried a kind of accent. Now, a rumor can arrive wearing the texture of evidence. A clip. A screenshot. A voice that sounds familiar enough to bypass your first doubt.
It changes what the body does first. Before you have decided what you believe, you may feel the quick lift of alarm or outrage. Because the clip looks complete, the feeling can attach to the story as if the sensation itself is proof.
A rumor is not just information, for it is a social event that asks for a reaction.
In the AI era, that request is sharper because the cost of producing convincing material has fallen. Synthetic media makes it easier to produce a receipt for an untrue claim, or to turn a vague suspicion into something that looks like a captured moment. The old work of rumor was imagination, the new work of rumor is fabrication with high production value.
The costs show up in places people do not expect.
One cost is literal. In May 2023, an AI-generated image appearing to show an explosion near the Pentagon went viral. Officials debunked it, but it still briefly rattled markets—a momentary shudder in the global nervous system before the claim was caught.
Another cost is time, and time is no longer a soft cost. When a believable clip spreads, journalists, officials, and ordinary group chats all spend attention disproving something that should never have existed. The rumor steals hours not by being true, but by being plausible enough to demand a response.
A third cost is the way rumors reshape trust itself.
In early 2022, a deepfake video circulated appearing to show Ukraine’s President Volodymyr Zelenskyy urging surrender. The clip was quickly challenged, but the point of such material is often not lasting belief—it is momentary confusion. It interrupts the sense that reality is shared. It teaches people that even the most recognizable faces can be made to say the wrong thing.
Confusion has downstream effects. It makes people suspicious of genuine footage. It also gives bad actors a second weapon: the ability to deny real evidence by claiming it is fake. Once fakes are common, denial becomes easier to sell.
Political rumors become more combustible because they trade in identity. In early 2024, New Hampshire officials investigated an apparent robocall that used AI to mimic President Joe Biden’s voice to discourage voting. A voice can be treated like authority even when it is only a performance. The rumor does not need to be elaborate, for it only needs to sound official enough to be repeated.
In Slovakia, deepfake audio surfaced in the days before an election, hitting during a pre-election media silence when rapid rebuttal was legally harder. And in Hungary in February 2026, a campaign video used AI to depict fictional scenes to suggest political stakes in an emotionally manipulative way. These are not just rumors, they are engineered feelings delivered through synthetic evidence.
The cost here is the deformation of public attention.
When people are repeatedly exposed to emotionally intense false content, their sense of proportion changes. Outrage becomes the default lens, cynicism becomes a protective posture. Some people disengage because believing anything feels exhausting—others double down on whatever source feels like home, because home is simpler than doubt.
There is also the financial and reputational cost that lands on individuals. In India, stock exchanges have warned investors about deepfake videos impersonating chief executives to give false stock recommendations. An authoritative face becomes a distribution channel for fraud. Even when a warning goes out, the false clip has already traveled—and the clean-up work rarely reaches everyone who saw the first version.
In workplaces, rumors carry a new kind of risk because synthetic media can impersonate authority. A well-publicized case in 2024 involved a deepfake video conference scam where an employee transferred large sums after seeing what looked like senior executives on a call. This is not a rumor in the old gossip sense, but it runs on the same fuel—a believable story delivered in a believable form, designed to trigger obedience before reflection catches up.
What makes AI-era rumors especially costly is that they exploit something decent in people.
Social life depends on shortcuts because no one can verify everything from first principles. Rumors take advantage of that dependency. They do not prove that people are foolish, for they prove that modern life requires trust to function.
There are legitimate benefits to these tools. Synthetic media can improve accessibility through translation or protect the privacy of those telling sensitive stories. The problem is not that generation exists, but that the boundary between play and proof has weakened.
The deeper cost is a new tax on attention.
Each time a convincing rumor spreads, people pay twice. First with the emotion it triggers, then with the work needed to undo it, or the numbness needed to survive it. Over time, that tax changes behavior. People become slower to help because they are unsure what is real. Or people become quiet because speaking into a noisy truth environment starts to feel pointless.
A society can live with disagreement, but it struggles when it cannot agree on what happened.
The way forward is partly technical, but it is also cultural. Rumors thrive where forwarding is effortless and being first is rewarded. They weaken when a pause becomes normal—when sharing without checking becomes slightly embarrassing. Those are not laws, they are habits. And habits, over time, become a kind of public infrastructure.
In the AI era, the most expensive rumors are not always the loudest ones. They are the ones that make ordinary trust feel naïve.
When that happens, even true stories arrive carrying extra weight. Not because they are less true, but because the mind has learned that plausibility is no longer evidence. The cost is a world that becomes harder to read—and a self that stays a little more guarded inside it.
The goal is no longer just to find the truth, but to protect the quiet intuition that allows us to recognize it.