A widely circulated image purportedly depicting one of the American airmen recently rescued by U.S. special forces from Iran drew millions of views this past week. It drew something else as well: a fact-check. The image, several accounts announced with evident satisfaction, is AI-generated. (Bravo, Inspector Clouseau.) Texas Governor Greg Abbott shared it. Major influencers amplified it. The fact-check lit up the reply threads.

I want to set aside the image and focus on the finger-pointer, because what the act of identification reveals is more interesting than the image itself.

Consider what was not contested. The airman’s rescue happened. The emotion expressed by millions of people who engaged with the image was genuine. No one who shared it claimed it was a photograph taken by a photojournalist embedded with the rescue team. Most people who encountered it likely experienced it the way they experience a commemorative illustration, as a visual token for something that actually occurred.

Now consider a cartoon. Suppose someone had drawn the same scene, soldiers in a helicopter, smiling, American flag in hand, in the style of a tasteful editorial illustration, the fact-checkers would have had nothing to say. The drawing would have traveled the same emotional circuit. The soldiers would have been the same soldiers. The rescue would have been the same rescue. The difference between the cartoon and the AI-generated image is purely procedural. The AI image was generated by a statistical model trained on visual data. The cartoon was generated by a human hand trained on visual instruction. In both cases, no camera was present. In both cases, the image is a representation, not a document.

Yuval Noah Harari argued in Sapiens that the human capacity for shared fiction, for constructing and inhabiting stories that are not literally true in a documentary sense, is the source of civilizational cohesion. The story of a rescued soldier, expressed in an image that was never a photograph, is doing exactly this work. It is binding a community around a shared recognition of something that happened and matters. The finger-pointer, by flagging the image’s generative provenance, is not adding epistemic content. The finger-pointer is asserting a procedural standard as a substitute for engaging with the story. The question “is this AI-generated?” has displaced the question “is this meaningful?” and the displacement is being performed as though it were a contribution to public discourse.

What the finger-pointer is actually doing is performing epistemic status. The detection requires no expertise, but deploying it produces the appearance of rigor: I saw through this, I identified the error, I am the one who knows. This is not fact-checking in any meaningful sense. Fact-checking interrogates claims and the claim here, that American airmen were rescued, is true. What is being fact-checked is the artwork.

This particular form of intervention will become self-obsolete. The precedent is already visible. When Photoshop entered the visual commons in the 1990s, “it’s been Photoshopped” carried the same accusatory charge the AI flag carries today. The charge faded because it became universal. Sharpening, cropping, color grading, exposure correction, skin retouching, background removal, these are now understood as the ordinary conditions of professional image-making, not deviations from it. Nobody pauses before a magazine cover to announce that the photograph has been post-processed.

As generative models improve and AI-generated imagery saturates the visual commons, the identification will carry decreasing signal. When every image could be AI-generated and many will be, announcing that a specific image is AI-generated will produce the same information as announcing that a specific sentence was typed on a keyboard.