Can You Trust a Photograph on the Internet?
A detective's guide to real images, edits and AI-generated pictures.
Imagine scrolling through a social media feed and seeing a photograph of an enormous shark swimming through the streets of a flooded city.
The caption says it really happened. The picture looks convincing. Lots of people have shared it.
Would you believe it?
A photograph can be powerful evidence of something that happened, but **seeing a picture isn't the same as knowing the story behind it**.
## Photographs have always needed context
Long before today's AI tools existed, photographs could be cropped, staged, edited or given a misleading caption. A real picture taken years earlier might be shared as though it had been taken that morning. A photograph from one country might be labelled as a completely different place.
Sometimes none of the pixels has been changed. The deception is in the words surrounding the image.
That's why the first question isn't necessarily 'Does this look real?' It might be 'Who took this, when and where?'
## What has AI changed?
Generative AI systems can create convincing images from text instructions. They can also alter existing pictures in ways that would once have required specialised editing skills.
AI imagery can be useful for art, diagrams and storytelling. Abby's Labs may use generated illustrations to help explain science. The important thing is to **label those images accurately**, rather than pretending they're documentary photographs.
The same tool can be used misleadingly if somebody presents an invented scene as a real event.
## Can your eyes always spot a fake?
Not reliably. Some artificial images contain strange details, but others do not. You shouldn't conclude that a picture is genuine just because hands, shadows and writing look normal.
And you shouldn't accuse somebody of using AI just because a photo looks unusual. Cameras, lenses, reflections and lighting can all create surprising effects.
## Become an image detective
Here's a better approach.
**Check the original source.** Who published the image first? Is it from a reputable organisation, a known photographer or an anonymous account?
**Check the date and location.** Does the image appear elsewhere with an earlier date or different description?
**Look for independent confirmation.** If the picture claims to show a major event, can trustworthy news organisations or official sources confirm it?
**Check the label.** Is it called an illustration, artist's impression, simulation or AI-generated image? Those labels can be perfectly honest - but they tell us the picture isn't an ordinary documentary photograph.
**Pause before sharing.** An image that makes you furious or frightened is exactly the kind of image worth checking before you pass it on.
## What if we still don't know?
Sometimes the evidence isn't available. It is completely reasonable to say, 'I'm not sure this picture is authentic.'
That's better than making a confident guess. Being curious also means being comfortable with uncertainty.
## Our own responsibility
When Abby's Labs uses a real photo, it should be correctly described. When it uses an illustration, it should say so. If an image is an artistic reconstruction, readers should understand that too.
Trust isn't built by making every page look impressive. It's built by helping people distinguish what is known, what is illustrated and what is still being investigated.
### Abby's Big Question
Would you rather see a less dramatic photograph with a trustworthy source, or an amazing-looking image without any evidence that it is real?