Fake News and Image Verification: A 5-Step Checking Guide
During an earthquake, flood, election or other major event, many of the images circulating on social media are old, taken somewhere else, or stripped of their context. This guide describes a five-step check you can run on an image before you share it.
1. Source: who shared it first?
Did the account that shared the image create the content itself, or take it from someone else? Look at when the account was created, what it has shared before and whether it could plausibly have been at the scene. Be careful with very new accounts that keep posting “breaking news” on unrelated topics.
2. First publication: has this image been shared before?
Reverse image search is the fastest way to catch a false context. If the image was shared years ago during another event, the new claim is most likely false. Sorting results from oldest to newest on TinEye helps you find when the image first appeared online. For videos, search the keyframes one by one.
3. Location: was it really taken there?
Do the signs, buildings, road markings and nature in the image match the claimed place? Compare with street view and satellite imagery. For details of the method, see our guide to finding a photo’s location.
4. Time: was it really taken that day?
- Weather: If it is raining in the image, did it rain in that city that day?
- Shadows: Do the direction and length of the shadows fit the claimed time?
- Season: Do the leaves on the trees and people’s clothes fit the claimed date?
- Archives: Look at an older version of the page the image appears on in a web archive.
5. Context: does the story hold together?
Even if the image is real, the claim written under it may be false. Check the event against reliable news sources and official statements. In Türkiye, independent fact-checking platforms such as Teyit and Doğruluk Payı examine widespread claims and publish their results; the claim you are looking at may already have been checked.
AI-generated images
AI-generated images are becoming more realistic. Still, they are often given away by some inconsistencies:
- Meaningless or garbled text (on signs, plates, T-shirts).
- Oddities in hands, fingers, ears and jewellery.
- Light and shadows and reflections that do not match each other.
- Illogical objects and architecture blending into each other in the background.
Has the image been manipulated?
Methods such as error level analysis (ELA) and clone detection can be used to tell whether part of an image was added later. These methods give clues but not definite answers; they can be misleading, especially on images compressed many times on social media.
Summary: ask before you share
- Who shared this first?
- Was this image online before?
- Was it really taken at the claimed place?
- Was it really taken at the claimed time?
- Is the story consistent with reliable sources?
The best way to build these skills is practice. Our geolocation missions train the hardest step of verification: the question of “where?”.