How to check an image before you believe it

Someone sends you a photograph. A politician doing something outrageous, someone you admire saying something cruel, a picture of a disaster or a too-good-to-be-true product. It looks completely real, and it makes you feel something straight away.

For most of the web’s life, “it looks real” was a fair guide. Convincing photographs were hard to fake, so a convincing photograph was usually a true one. That rule has quietly stopped working.

Why “it looks real” isn’t proof any more

AI image tools are trained on enormous collections of pictures — millions of them — until they have learned what things look like well enough to produce brand-new images from a short written description. Ask one for “a flooded high street with a shark swimming down it” and you’ll get a photograph of exactly that, lit and shadowed like the real thing, in seconds.

That’s genuinely useful for art and mock-ups. It also means anyone can now manufacture a convincing picture of an event that never happened, a person who was never there, or a product that doesn’t do what the picture shows. An image can be flawless and still be fiction. So the useful question is no longer “does this look real?” It is “where did this come from, and why am I seeing it?”

Four checks that still work

You don’t need to be a detective, and you don’t need to out-spot the fakery. You need a few seconds and a couple of habits.

1. Start with the source, not the picture. Before you react, look at where it came from. Who posted it, and do they have a reason to want you angry, frightened or excited? A photo from a news organisation that puts its name to its work is a very different thing from an image with no clear origin, forwarded by a stranger or an account you’ve never heard of. If you can’t find where a picture actually started, treat it as unproven.

2. Do a reverse image search. This is the most useful skill here, and it takes about ten seconds. A reverse image search looks for everywhere else a picture has appeared online. On a computer, drag the image into Google Images or a free tool like TinEye; on a phone, most browsers let you search an image you are looking at. It will often tell you a picture is years old, from another country, or already flagged as fake — before you have passed it on.

Try it yourself: Google — search with an image · TinEye

3. Be most careful with images that rush you. The pictures designed to fool you are usually the ones built to make you feel something immediately — fury, fear, awe, the urge to buy now. That heat is doing a job: it is there to hurry you past the part of you that would otherwise stop and check. When an image makes you want to act this second, that is the moment to slow down.

4. See if anyone else has it. If something genuinely important has happened, more than one source you already trust will be carrying it. A dramatic image living in only one corner of the internet, with no established outlet touching it, is a reason for doubt rather than a scoop.

What about the tell-tale signs?

You may have heard you can spot AI images by odd hands, garbled text on signs, or reflections that don’t line up. Sometimes you can, and a glance never hurts. But treat these as a bonus rather than your defence: the tools improve every few months, and the giveaways that work today may be gone by Christmas. A picture from nowhere is suspect whether or not the hands are right — which is why the source check is the one worth building into a habit.

Better questions, not sharper eyes

The reassuring part is that protecting yourself doesn’t depend on beating the technology at spotting its own mistakes, a contest you would slowly lose. It depends on one small, durable habit: where’s this from, and why am I seeing it now? Ask that, spend ten seconds on a reverse search when it matters, and your own judgement is back in charge of what you believe. You don’t need sharper eyes. You need better questions.