For image verification, start with source and reuse checks. For video verification, inspect several frames and the sequence around them. TinEye and Google Lens help locate visual matches; InVID supports keyframe work; ExifTool inspects file metadata. Maps, imagery, and sun-position tools address place and time. Each answers a different question.
Choose a tool for the question
Has this image appeared elsewhere?
Tools to consider: TinEye and Google Lens. A match is not necessarily the original; no match does not prove originality.
Which parts of a video can I search?
Tool to consider: InVID Verification Toolkit. A keyframe omits motion, sound, and surrounding events.
What metadata is in the available file?
Tool to consider: ExifTool. Metadata may be edited, stripped, inherited, or unrelated to capture time.
Is creation or editing provenance available?
Tool to consider: Adobe Content Authenticity Inspect. Credentials do not establish the truth of the depicted event; absence is inconclusive.
Does the proposed place and time fit?
Tools to consider: Google Earth, Mapillary, and SunCalc. A place match does not independently verify the date or caption.
A practical sequence for a viral image or short clip
- Define the claim. Separate the alleged location, date, people, event, and edits. Avoid a single vague “real or fake” question.
- Preserve the source. Record the public URL, caption, uploader context, visible time, and your collection time. Keep the original file separately from working copies. SingleFile can preserve a readable page; Browsertrix supports scoped web collections. Neither guarantees every video or dynamic element is captured, so verify the saved result.
- Search multiple views. Search the full image, useful crops, and several distinct video frames. Inspect candidate source pages rather than trusting a thumbnail.
- Check file and provenance evidence. Inspect available metadata and credentials. Record both their value and why the file is close enough to the source for that value to matter.
- Test place and time independently. Compare landmarks, roads, terrain, imagery dates, weather, shadows, and contemporaneous reporting. Document conflicting clues.
- Write a bounded conclusion. State what is confirmed, contradicted, plausible, or unresolved, with the evidence supporting each part.
How to interpret reverse-image dates
TinEye’s “first found” date records when its crawler discovered an image, not when the photograph was taken or first published. An older matching page can challenge a “taken today” caption, but the earliest search result does not automatically identify the creator. See TinEye’s date explanation.
How to verify a short video
Extract frames around meaningful changes: a sign, building, transition, or identifiable landmark. Search several frames because a title card or generic sky view may reveal little. Compare the sequence before and after the claimed event when it is lawfully available. Audio and subtitles should be examined separately; a genuine frame can accompany an altered soundtrack or misleading caption.
The InVID project documentation explains its keyframe and inspection workflow. Platform support and individual services can change. Confirm the current extension and relevant service before relying on an older tutorial, and check processing arrangements before uploading sensitive media.
Metadata and AI-detector limits
ExifTool reads a wide range of file metadata. A camera or timestamp field is a clue, not proof: exports, screenshots, editing software, and intentional modification can alter it. Preserve the original before changing metadata.
An AI-detector score should not settle an authenticity claim. Seek positive evidence such as source history, corroborating observations, and documented creation context. Content Credentials can describe available provenance; they do not replace verification of the claim. Do not penalize authentic material solely because credentials are absent.
Prepare a publication copy
Preserve the evidence separately, then remove unnecessary personal details from the copy you will share. Blur Face is a redaction utility, not a verification tool. Automatic detection can miss a face or plate, so review the export, captions, audio, filenames, and metadata before publication. Protect bystanders, sources, and sensitive locations.
Methodology and further reading
Official documentation reviewed September 26, 2026. Tool roles are based on documented capabilities; this page does not claim a new hands-on test or an accuracy benchmark. For an extended checklist, use Verify an Image or Short Video, TinEye vs InVID vs ExifTool, and the geolocation collection.
