10 Websites That Help You Verify Almost Anything Online

Main ideas & lessons (what the video teaches)

• Verification is about evidence, not “magical truth.” Each tool can provide useful clues, but none is definitive on its own.
• Start from the right question for the right tool. The “secret” is matching tools to the type of claim (image vs. video vs. webpage vs. location vs. time).
• Don’t stop at the first answer. Tools can be wrong or incomplete; stronger conclusions come from multiple independent sources agreeing.
• Investigate discrepancies rather than assuming dishonesty. Metadata or tool outputs can be misleading due to editing, platform stripping data, reposting, or normal processing.
• The internet leaves traces. Files, maps, satellite imagery, search engines, and archives retain history that can be used to reconstruct events or refute/location/date claims.

---

Methodology (explicit workflow given in the video)

General verification principle

Use tools to collect independent evidence, such as:

• file metadata
• visual/sign details
• street geography (maps + ground imagery)
• satellite/geographic patterns (aerial imagery)
• image/video history across the web
• archived page history

Example step-by-step workflow (used to illustrate the approach)

When a suspicious post appears (photo + claim of extraordinary event at a specific location and time):

1. Check file metadata

   • Use metadata inspection to look for:
     • creation timestamps
     • device/software information
     • image dimensions
     • possible GPS coordinates (if present)
   • Interpret carefully:
     • metadata may be missing, stripped, changed, or edited
     • metadata inconsistencies are leads for deeper checking, not proof by themselves

2. Search inside the photo for identifying text/logos

   • Use Google Lens to crop and search small details (signs, logos, storefront text, distinctive architecture, symbols).
   • Use results to identify businesses/brands/models/locations; refine separate questions by cropping different regions.

3. Verify street-level geography

   • Use OpenStreetMap to compare:
     • roads, footpaths, rail lines, building outlines, intersections, land-use boundaries, landmarks
   • Focus on geometry rather than “does it look like X city?”
   • Also consider edit history/corroboration if available.

4. Corroborate with ground imagery

   • Use Mapillary if metadata is washed out by reposting/screenshotting/reuploading.
   • Look for matches to background details (walls, gates, utility poles, ridges).
   • Use “older imagery” and time travel / comparisons when supported to see whether specific features existed then.

5. Confirm larger geographic patterns from above

   • Use Google Earth to compare aerial patterns:
     • coastline shapes, river angles, road layouts, airport/runways, industrial geometry, neighborhoods/street grids, mountain shadows, field patterns
   • Important misconception corrected:
     • Google Earth is not live; imagery is collected over time and may be a composite of dates.
   • Use measurement tools, coordinates, and (where available) historical imagery.

6. Verify timing over long periods

   • Use Google Earth Time Lapse to check large-scale changes across decades.
   • Limits:
     • not suitable for “what happened last Tuesday”
     • imagery dataset ends in 2022
     • cloud cover, seasons, and image quality affect results

7. Find whether the same image appeared earlier

   • Use TinEye (reverse image search).
   • Identify older copies, crops, higher-resolution versions, edited variations.
   • Caveat:
     • “Oldest result” may mean earliest occurrence in TinEye’s index, not necessarily the true original source.

8. If it’s a video, extract still clues

   • Use InVID:
     • extract key frames into representative still images
     • run reverse searches on key frames (often more effective than searching the full moving clip)
     • use additional forensic/magnification/metadata/image analysis tools as needed
   • Warn against a misconception:
     • there’s no single “fake video” button; artifacts can come from compression, normal processing, screenshots, overlays, etc.

9. Confirm or recover what web pages used to say

   • Use Wayback Machine:
     • view snapshots over time (wording changes, deleted pages, old product/policy statements)
     • use “save page now” when relevant, noting archives may be incomplete or broken for dynamic/Javascript-heavy pages
   • Still interpret changes—archives document “what,” not always “why.”

10. Stack evidence

    • Combine independent lines of support (file data + ground + aerial + image history + page history).
    • If sources agree, confidence increases.
    • If they conflict, that conflict often reveals the most interesting lead.

---

Tool-by-tool main concepts (the “10 websites/tools” listed)

10) Metadata2Go (file metadata inspection)

• Extracts metadata from compatible digital files.
• Can reveal:
  • device used, image dimensions, timestamps, software info
  • sometimes GPS coordinates
• Key lessons:
  • metadata can be removed/changed
  • platforms may strip metadata
  • missing/odd metadata doesn’t prove deception; treat as a lead

9) OpenStreetMap

• Collaborative map with dense, sometimes subtle geographic features.
• Verification focus:
  • compare road geometry and surroundings against claimed location
• Also notes:
  • edit history can help in some cases
  • info can be incomplete/outdated/wrong → corroborate

8) Mapillary

• Crowd-sourced street-level imagery (from phones/dash/action/360 cameras).
• Verification use:
  • find matching walls/gates/poles/ridges/road segments behind a claim
• Supports:
  • exploring multiple captures
  • older imagery and time-related comparisons where available
• Limitations:
  • coverage gaps; variable image quality
  • randomness can still help when unrelated contributors captured the same place

7) Google Earth

• Aerial/pattern-based location verification.
• Verification value:
  • roads/coastlines/river bends/airport layouts/industrial geometry can become obvious from above
• Key correction:
  • not live; imagery is time-collected/composited
• Can use:
  • coordinates, measurement tools, 3D views, historical imagery (supported areas)

6) Google Earth Time Lapse

• Tracks long-term landscape change using decades of satellite imagery.
• Best for:
  • whether large developments/expansions/removal of forests/coastline shifts occurred over years/decades
• Limits:
  • not for checking “last Tuesday”
  • dataset timeframe ends in 2022
  • cloud/season/image-quality issues affect results

5) TinEye (reverse image search)

• Searches by image rather than text.
• Finds:
  • matching/modified versions across the web
  • older copies, crops, higher resolution, edited variants
• Helps catch “cheap tricks”:
  • genuine image used with wrong date/country/event/caption
• Important caveat:
  • “oldest” = earliest time found in TinEye index, not necessarily the true first origin

4) Google Lens

• Searches inside an image by cropping recognizable details.
• Useful when:
  • most of the photo is generic, but a sign/logo/text/symbol is specific
• Can also:
  • find visually similar images/pages
  • isolate multiple regions to ask separate questions (building, sign text, company logo, vehicle model, monuments)

3) InVID (We Verify)

• Designed for video verification challenges (reposts, compression, screen records, cropping, burned captions/music).
• Key feature:
  • extract key frames (turn video into searchable still images)
• Why it matters:
  • reverse image search works much better on still frames than on the whole 30-second video
• Additional tools:
  • magnification, metadata inspection, image analysis, forensic examination (with some access restrictions)
• Warning:
  • no automatic “fake detection” certainty; artifacts can come from normal processing

2) Wayback Machine

• Internet Archive’s web page history tool.
• Lets you:
  • view archived snapshots by date
  • compare wording/product/policy changes
  • recover deleted or rewritten pages
  • request future snapshots (“save page now”)—with limitations
• Limitations:
  • not complete; login-protected or JS-heavy pages may fail; images/videos may be missing
  • archive shows changes but not necessarily the reason

1) AFP Digital Verification Toolbox (organized workflow + tool selection)

• Presents a “toolbox” approach rather than a single magic site.
• Purpose:
  • organize tools for tasks like reverse image search, geolocation, metadata analysis, archives, social media investigation, and AI-related verification
• Core message:
  • verification is workflow + stacking evidence, not relying on one tool alone

---

Speakers / sources featured (as mentioned in the subtitles)

• AFP (Agence France-Presse) – “AFP digital verification toolbox”
• OFP / “OFP, the International News Agency” (as referenced in the video)
• Internet Archive (operator of the Wayback Machine)
• TinEye (tool/source)
• Google (Google Earth, Google Earth Time Lapse, Google Lens)
• OpenStreetMap (platform)
• Mapillary (platform)
• InVID (We Verify) (tool)