How do journalists verify photos and videos from social media?
Through a set of open-source verification techniques that have become a specialism in their own right, largely because unverified imagery spreads faster than newsrooms can check it.
The core questions any verification process answers: where was this taken, when, who took it, and is it what it claims to be?
The main techniques:
Reverse image search. The first step and frequently sufficient. Running an image through search tools reveals earlier appearances — and the most common form of misinformation is not fabrication but recontextualisation: genuine footage from a different place or a previous year presented as current.
Geolocation. Matching features in the image to satellite and street-level imagery — building shapes, road layouts, signage, terrain, distinctive structures. This can pin a location precisely, and is the technique that has established facts in numerous conflict investigations.
Chronolocation. Establishing time from shadows — length and direction give sun position, which combined with location gives date and time — plus weather records, vegetation state, and visible construction.
Metadata, where it survives. Note that most platforms strip EXIF on upload, so its absence proves nothing and its presence should be treated cautiously since it can be edited.
Source contact. Asking the uploader directly whether they filmed it, on what device, and what was happening just outside frame — which fabricators frequently cannot answer.
Cross-referencing against other footage of the same event from different angles.
Technical analysis — inconsistencies in lighting, shadows, reflections and edges.
The AI complication. Generated imagery has made visual inspection far less reliable, and detection tools are inconsistent. The response has been to place more weight on provenance — where did this come from and can it be traced — than on examining the image itself. Content authentication standards attaching verifiable origin data are being deployed for this reason.
Organisations doing this well: Bellingcat, and newsroom verification desks at major outlets.
"We cannot independently verify" in a report is meaningful and should be read literally.