Journalist's guide: verifying viral images before publishing

V — View the metadata. Run EXIF through ExifTool or ScanTrace. Check for software tags, coordinates, original timestamp.

E — Evaluate forensically. Upload to ScanTrace. Record the verdict (REAL / AI_GENERATED / INDETERMINATE) and confidence score.

R — Reverse search. Google Lens, TinEye, Yandex, Bing Visual Search. Yandex is famously the best for faces and European subjects.

I — Identify and contact the source. The person or agency that first posted the image. A 60-second DM often beats any detector.

F — Find corroborating evidence. Cross-reference with news agencies, official accounts, eyewitnesses, weather data, satellite imagery.

Y — Yes or no? Only publish when the five steps align. One red flag = hold.

Assign one verification officer on every shift. Give them a checklist (the protocol above) and 10 minutes of budget before any breaking-news image goes live. Larger newsrooms — BBC, Le Monde, El País — have dedicated OSINT desks with 3–8 journalists. Smaller outlets can rotate the role weekly.

A Midjourney image of Pope Francis in a white puffer jacket went viral on Reddit in March 2023. Within 48 hours it was shared uncritically by hundreds of mainstream outlets. Three checks would have caught it: (1) absence of EXIF, (2) reverse search matching Reddit as the earliest source, (3) lack of photographer credit in any Vatican pool. A 90-second protocol would have prevented a global journalistic failure.

Bellingcat, the gold standard in OSINT journalism, publishes its verification methods openly. Their three pillars: transparency (show your work), replicability (anyone can follow the same steps), and multiple independent sources (never rely on a single signal). Every paragraph in this article maps to one of those pillars.

ScanTrace fake photo verifier, Google Lens, Yandex Images, TinEye, InVID-WeVerify plug-in, ExifTool, SunCalc, Mapillary, Wayback Machine, CrowdTangle archive, Bellingcat's Online Investigation Toolkit.

A trained desk runs the full protocol in under 7 minutes. That is less than the average editorial meeting. The reputational ROI of not publishing a fake is massive; the cost of one ScanTrace scan is cents. Open a free ScanTrace account and integrate it into your newsroom workflow this week.

How long should verifying a viral image take?

A trained journalist running the 5-step protocol can verify a single image in 4–7 minutes. Under breaking-news pressure the forensic + reverse search steps alone (2 minutes) already filter 80% of fakes.

Which tools are essential for newsroom verification in 2026?

A forensic detector (ScanTrace), a reverse image search (Google Lens + Yandex), an EXIF reader, a geolocation tool (Google Earth + Mapillary) and a weather/sun position checker (SunCalc). Bellingcat's Online Investigation Toolkit lists dozens more.

What if the image comes from a verified Twitter/X account?

Verified accounts are not a guarantee. Blue checks are purchased. Run the protocol anyway. Bellingcat's rule: 'Every image, every time, no exceptions.'

Can I publish if ScanTrace says REAL but reverse search finds nothing?

Yes, but add a disclaimer: 'Image could not be independently cross-referenced, published under the responsibility of source X.' Transparency is always better than silent confidence.

What is the legal risk of publishing a fake image?

In the EU the Digital Services Act imposes content-moderation duties that extend to professional publishers. National press laws also recognise 'duty of verification' as a liability shield — if you didn't verify, you can be sued for damages caused by the fake.