Emergency fact-checking: checking photos and videos in real time
Every day, thousands of shocking photos and videos appear on social networks and messengers, claiming to be exclusive footage from the scene. How to distinguish real content from a skillful fake or archival footage when every minute counts? This instruction is your algorithm of actions for emergency fact-checking, which will help you quickly and effectively verify visual information before it is published.
1. Check for archival status (reverse image search)
An old post presented as fresh is a classic sign of a fake. Therefore, before working with photo and video material that is declared as new, it is worth making sure that it really has not appeared on the Internet before:
– Use reverse search services or “image search” (Google Images) — the earliest publication dates will indicate the original source.
2. Find the original source. Subsequent reposts often distort the facts, so it is best to use information from the original source, that is, find the very first publication of this material on the Internet. Only the original source can provide the exact context: where, when and under what circumstances the photo was taken.
3. Monitoring social networks and analyzing the context. Social networks are a living chronicle of events. Here you can find eyewitness accounts, additional angles, comments that will confirm or refute the information.
– Search by keywords (geolocation, names, event essence);
– Use advanced search within platforms, filtering by date and popularity;
– Search by hashtags.
It is important to note that information from social networks is not reliable by default, so accounts/channels distributing content always require additional verification.
4. Analyze metadata and statistics. Metadata (EXIF) of a digital file can contain valuable information: date and time of creation, camera type, geolocation. Read more about how to analyze metadata in the instructions.
It is important to remember that data can be falsified, so it is worth considering it only as one of the confirmation factors. In addition, file metadata can be deleted — accidentally or due to the specifics of the algorithms. In this case, analysis of publication statistics comes to the rescue.
5. Check for traces of generative AI. Neural networks have learned to create extremely realistic images and videos. It is important to be able to identify their artifacts:
– Look for anomalies in textures (especially hair, skin, water), illogical shadows, absurd details (for example, strange fingers, teeth, watches without hands).
– Use specialized detectors to identify generated AI content (Deepware Scanner).
The key principle: any information requires double-checking, and a few minutes, even under time pressure, is a small price to pay for reliability and a preserved reputation.