Guide · Updated · By William Lodge
How to Check If an Image Is Fake, Edited, or Out of Context
A step-by-step image verification workflow: reverse image search, metadata, forensic analysis, and geolocation clues — plus the limits of each method.
Most misleading images aren't sophisticated fakes — they're real photos shared with a false caption, from a different place or year. That's why image verification starts with provenance (where did this first appear?) before forensics (was it edited?).
Step 1: Save the best copy you can find
Social platforms recompress and strip images. Find the highest-resolution version, save it, and keep the original file untouched — work on copies. Note where and when you found it.
Step 2: Reverse image search
Run the image through several reverse search engines, because each has a different index: Google Lens, Bing Visual Search, Yandex Images, and TinEye. TinEye is especially useful for sorting results by oldest first. You're looking for:
- The earliest appearance online — often the real origin.
- Versions with a different caption, date, or location.
- Higher-resolution or uncropped versions that show more context.
Crop to a distinctive part of the image and search again if full-image searches return nothing.
Step 3: Check metadata (EXIF)
Original photo files can include camera model, capture time, software used, and sometimes GPS coordinates. Forensically shows metadata in the browser. Treat metadata as a lead, not proof: most social platforms strip it, and anyone can edit it.
Step 4: Look for signs of editing
Forensically includes several forensic views. The most useful for beginners:
- Error Level Analysis (ELA) highlights areas compressed differently from the rest of the image — a pasted-in object can stand out.
- Clone detection finds regions copied from elsewhere in the same image.
- Noise analysis reveals areas whose noise pattern doesn't match their surroundings.
These tools are easy to misread. Recompression, resizing, and text overlays all create artifacts. Use forensic results to decide where to look harder, not as a verdict.
Step 5: Verify the location
If the image claims to show a specific place, check it against satellite and street-level imagery. Match several independent features — building shapes, road layouts, signs, mountains, power lines — rather than one. Shadow direction and length can also help estimate time of day. Our geolocation tools and the Image Research OSINT guide go deeper.
Step 6: Check for AI generation
AI-generated images often have tell-tale problems: inconsistent text and signage, malformed hands or jewelry, mismatched reflections, and backgrounds that dissolve into nonsense. Detection tools exist but are unreliable on their own; reverse search showing no earlier source, combined with visual artifacts, is a stronger signal than any single detector score.
Report with confidence levels
Write up what you confirmed (earliest source, matched location) separately from what you suspect (possible editing). Keep links to every source and archive them — see how to archive a web page as evidence.
Privacy note: don't upload private or sensitive images to third-party services. Browser-based tools that process locally are safer for sensitive material.
You are responsible for authorization and the laws that apply to your research. OSINTCompass lists tools for lawful, ethical public-source research.