What "verification" means
Open-source intelligence (OSINT) analysts — researchers who work from publicly available material rather than classified sources — have become central to conflict reporting. Groups like Bellingcat, a non-profit funded by donors, pioneered methods now used across newsrooms. But verification is not a single trick. It is a checklist, and footage fails the checklist far more often than it passes.
The four questions, per Bellingcat's published methodology:
1. Where was this filmed?
2. When was this filmed?
3. Is the file itself authentic?
4. Has this footage appeared before?
1. Geolocation: matching the ground
Geolocation means matching visual clues in the footage — buildings, road layouts, terrain, signage — against satellite imagery, street-level views, and maps. Google Maps and Google Earth Pro (free) are the workhorses.
A striking technique uses shadows. The length of an object's shadow, compared with the object's height, reveals the sun's elevation — and the sun's position is a function of place and time. Bellingcat explains the concept through the subsolar point: at any moment there is one point on Earth where the sun is directly overhead, and locations at equal distance from it form a ring of equal sun angle. Given a date and time, an object's height, and its shadow length, analysts can draw a ring of possible locations — and Bellingcat's Shadow Finder tool can narrow the search from the entire globe to a handful of countries.
Shadows can also be checked manually with tools like SunCalc. Ideal conditions: the date and time are known, the object and its shadow sit roughly at right angles to the camera, the shadow falls on a level surface, and there is no fisheye distortion. Multiple images taken at different times produce overlapping rings that narrow the candidates further — in one Bellingcat example, two candidate locations (Cameroon vs. Croatia) were reduced to one when the inland option was eliminated, and the remaining coastal site near Split was confirmed via street view.
2. Chronolocation: pinning down the time
Even correctly located footage may be old. Analysts check the claimed date against weather records — a photo allegedly taken on a sunny 1 June 2000 in Amsterdam can be checked against that day's actual weather — and against timezones, converting upload and post timestamps to UTC (Bellingcat recommends the Uniform Timezone Extension for this).
3. File forensics: is it real?
Metadata (EXIF data) embedded in image files can reveal camera, settings, and timestamps — but it may also be inaccurate or stripped, so analysts treat it with caution and cross-check it with tools like Metadata2go or EXIFmeta. For video, Bellingcat's starter pack recommends the InVID browser extension, which helps detect and authenticate newsworthy video, including checking for signs of editing.
4. Provenance: has it been seen before?
Reverse image search is the essential first step. Old footage recirculated as new is one of the most common forms of conflict misinformation. Analysts run frames through Google Lens and multi-engine tools like RevEye (which queries Google, Bing, Yandex, and TinEye). If the footage already existed online under a different claim, the investigation effectively ends there.
Analysts also archive everything — via the Wayback Machine, Archive.today, or Bellingcat's autoarchiver — because original posts are frequently deleted.
The limits
Bellingcat's own guidance carries an important warning: if a photo turns out to have been taken somewhere completely different from what was claimed, be cautious about the other information attached to it. Verification is about the footage, not the narrative — and a verified location does not make every claim in the caption true.
