On May 11, 2026, a photograph of what looked like a page from a computer science textbook spread across Reddit and X. The page appeared to contain a paragraph phrased the way a chatbot answers a question, as if an AI-generated response had been pasted into the manuscript and nobody had edited it out. The reaction was swift: thousands of comments lamenting the state of publishing, education and human attention.
What almost nobody in those threads could establish was whether the page was real. No one named the book, the author, the publisher or an ISBN. Several commenters pointed out that current image models can produce convincing pages of printed text, and argued that the photo itself was likely synthetic. Coverage of the episode, including a Startup Fortune analysis published the same day, treated the image as unverified and noted that it could be genuine, staged or entirely generated.
Either way, the episode is a small illustration of a larger shift: a picture is no longer evidence of anything on its own.
Why fakes are harder to spot
A few years ago, manipulated photographs usually left traces that a careful viewer could find: mismatched lighting, warped backgrounds, hands with the wrong number of fingers, garbled lettering. Image generators have improved on all of these fronts. Rendering legible text, once a reliable giveaway, is now routine for leading models. Lighting, skin texture, fonts, camera angles and facial expressions can all be produced convincingly enough that an ordinary viewer scrolling on a phone has little chance of telling a generated image from a photograph.
The same applies increasingly to video and audio. Cloned voices and synthetic video are already used in scams, and live video calls can be faked in real time with enough preparation.
The consequences reach beyond social media
The textbook image, if it was fabricated, did little harm beyond wasting a day of online outrage. The same techniques, applied to politics, finance or the courts, raise harder problems: fabricated statements by public figures, fake evidence in disputes, forged documents used in fraud, and manufactured outrage designed to shift public opinion. In each case, the fake does not need to survive scrutiny for long. It needs only to be believed by enough people before anyone checks.
The deeper issue is that the human habit of trusting what we see is not built for this environment. People react to an image emotionally within seconds. Verifying it takes minutes or hours, and most never attempt it.
Skepticism as a basic skill
Researchers and media literacy educators have argued for several years that the appropriate response is not to distrust everything but to adopt a verification habit, especially for content that triggers a strong reaction. Practical steps include asking who first posted an image and whether that account has a track record, checking whether established news organizations or the people supposedly involved have confirmed it, looking for the original source rather than a screenshot of a screenshot, and waiting before sharing.
In the textbook case, those questions would have surfaced the problem immediately: no one could say which book it was. That absence of a traceable source is, increasingly, the most useful signal available. The internet was built as a way to distribute information. It is now also a way to distribute convincing imitations of it, and the burden of telling the two apart has moved onto the reader.
