On 22 January 2026, the White House posted an image on its official social media account showing federal agents leading away the civil rights lawyer Nekima Levy Armstrong from a protest against Immigration and Customs Enforcement, or ICE. In the White House version, Armstrong appears to be crying uncontrollably, her face contorted with anguish.

Only minutes earlier, US Secretary of Homeland Security Kristi Noem had shared a different image of the same moment — one in which Armstrong looks composed and calm. Comparisons by several news outlets quickly revealed that the image published by the White House had been manipulated.

When journalists asked about the authenticity of the image, the White House did not deny that it had been altered. Instead, the press office referred reporters to a post by deputy communications director Kaelan Dorr, who brushed aside the question with the line: “Enforcement of the law will continue. The memes will continue.”

Memes continue. Do not expect a correction. Do not expect shame. Do not expect being caught to change anything. The doctored image was never meant to serve as evidence. The point was to humiliate an opponent and move on.

When the media seizes on the forgery and turns it into an exposé, that is not a defeat for the manipulator but a victory. In this set-up, journalism begins to feel embarrassingly naïve, even stupid. It is trying to play by the rules of truth on a field where the other side long ago abandoned the very idea of the image as evidence — or, indeed, of evidence itself.

When Brooklyn Beckham recently accused his family in public of sabotaging his marriage, the audience did not wait around for proof. It produced its own: realistic AI-generated videos and images in which Victoria Beckham dances “inappropriately” at the wedding. No real-life counterpart to this synthetic material exists.

The videos were shared, liked and remixed. Many viewers seemed to understand that they were fake, but that did not matter. The content was plausible enough to satisfy the audience’s curiosity, and that was enough.

Finland, too, has produced its own example of what keeps this ecosystem going. In December 2025, Yle’s verification team traced a network of Facebook pages churning out synthetic “facts” about Finland. According to the network, Finnish classrooms are dome-shaped buildings with roofs made of moss, and outside shops there are electrically heated posts where dogs can keep warm while their owners are shopping.

The images and claims were AI-generated. The pages are run largely from India and Pakistan, and they make money. In 2024, Meta, Facebook’s parent company, paid more than $2 billion in total to Facebook content creators worldwide.

The material is aggressively innocuous: cheerful Nordic magical realism engineered to spread. Yet its broader effect is anything but harmless. When platforms reward the mass production of plausible-sounding nonsense, they strengthen an information environment in which the boundary between fact and fiction begins to blur.

The spread of synthetic content has also created a new kind of advantage for liars. Legal scholars Bobby Chesney and Danielle Citron call it the liar’s dividend: the benefit lies not in creating a convincing fake, but in making falsification itself possible.

Once the public accepts that anyone can easily fake voices, images and videos, a shameless actor can brand any inconvenient piece of evidence as suspect. He does not need to prove it false. He can simply claim it was made with AI.

In September 2025, the Associated Press reported on a viral video in which someone appeared to be throwing objects from a White House window. President Donald Trump dismissed the video as AI-generated, even though the White House press team had previously said the footage was authentic.

“If something happens that’s really bad, maybe I’ll have to just blame AI,” Trump said.

Set against this, the doctored image of Levy Armstrong takes on a second meaning. It teaches the public by example that images can be distorted and therefore shrugged off. Especially when something really bad happens.

The philosopher Hannah Arendt understood where this road leads. Writing more than 70 years ago about Europe’s slide into totalitarianism, she observed that the most useful citizen for such regimes is not the convinced zealot but the exhausted sceptic: a person for whom the difference between truth and falsehood no longer means anything.

Propaganda’s achievement is not to plant a single lie, but to erode the ground on which truth stands — until correcting falsehoods begins to feel beside the point, and exposing them changes nothing.

This fatigue is fed by the way so many people now consume their news. In the social media feed, journalism is just one voice among many. Official statements, rumours, parody and forgeries all compete there for attention.

The danger is that, in order to survive in this environment, journalism begins to resemble its rivals more and more. In Finland, outlets including Demokraatti and Verkkouutiset have resorted to AI illustrations.

The journalists’ trade magazine Press Gazette recently published a list of 50 purported experts whose comments had been quoted more than 1,000 times across British newspapers, magazines and digital outlets, even though no reliable evidence of their existence could be found.

At the same time, AI-generated slop is already mimicking the form of news itself. The result is a flat, undifferentiated stream of content that we watch with weary shrugs: maybe true, maybe not. The liar gets his dividend.

But the game is not yet lost. Events in Minneapolis in recent weeks have been a reminder that photos and video still retain the power to burst the bubble of cynicism and demand accountability for acts those in power would much rather sweep under the carpet.

This article was first published in Journalisti on 28 January 2026 under the title “Totuudenjälkeisessä ajassa vääristelystä kiinnijääminen on voitto”.