The doctored photograph is now the workhorse of online misinformation: cheap to make, hard to unsee, and engineered to be shared before it is checked. Yet most fakes are not masterpieces. They fail in predictable places — light that falls inconsistently, shadows that point the wrong way, edges that smear where two images were joined. Learning to look for those failures takes minutes and works on the majority of viral fabrications.
Start with physics. In any real scene, light comes from identifiable sources, and every shadow agrees about where they are. A person whose face is lit from the left while the building behind them is lit from the right is standing in two photographs at once. Reflections are even less forgiving: eyes, windows and glossy surfaces preserve the scene in miniature, and forgeries routinely forget to forge the reflections. Guides like this one on realnewsanalysis.com exist because the checks are learnable skills, not talents.
The thirty-second reverse search
The single most powerful test requires no expertise at all: search for the image itself. Most ‘shocking’ photos are real photographs with false captions — an old disaster repackaged as today’s, a film still presented as news. A reverse image search reveals earlier appearances and original contexts in seconds. Verification organisations such as Bellingcat have built an entire methodology on this principle, cross-referencing landmarks, signage, weather records and shadows against the clock to establish when and where an image was truly made.
Metadata can help when it survives — many platforms strip it — but the absence of metadata proves nothing either way. More reliable are the small absurdities generative tools still produce: hands with the wrong count of fingers, text that dissolves into gibberish, jewellery that melts into skin, straight lines that wobble through objects.
Slow down at the spike of emotion
Institutions are building the habit into the product. Some newsrooms now publish verification notes alongside contested images, showing the geolocation and source checks performed, and platforms experiment with context labels on viral media. The direction is toward a web where provenance travels with the image — embedded capture data from cameras, signed at creation — so that the check becomes a glance rather than an investigation.
Until that infrastructure matures, the reader’s toolkit remains craft and scepticism: the physics of light, the archaeology of earlier uploads, and the discipline of the pause. Professional fact-checkers describe their advantage over the public as nothing more exotic than taking the minute the sharer did not. The minute is available to everyone. It is the whole toolkit, and it works.
The final check is psychological. Fabrications are engineered to produce an emotion spike — outrage, vindication, fear — because emotion disables scrutiny and powers sharing. Fact-checkers’ simplest advice remains the strongest: if an image makes you feel something powerful instantly, that is the moment to run the checks, not to forward it. Sixty seconds of scepticism is the tax misinformation cannot afford for you to pay.
