Skip to content

Citizen Evidence Lab YouTube Data Viewer

FreeFirst-line

Publisher: citizenevidence.org | Type: Non-detector | Cost: Free

TL;DR

A free web tool from Amnesty International's Citizen Evidence Lab. Paste a YouTube URL and the tool returns the exact upload time and every video thumbnail: the two anchors that let a fact-checker establish how old a clip really is and cross-check whether earlier copies of it exist elsewhere. Narrow scope, high reliability, no upload of source files.

What it does

The YouTube Data Viewer takes a YouTube video URL and returns two outputs: the precise upload timestamp (down to the second, not just the date YouTube displays publicly) and the full set of video thumbnails YouTube generates for the clip. The thumbnails matter operationally because they can be reverse- image-searched to locate earlier instances of the video; a clip claimed as freshly recorded that surfaces a 2019 thumbnail match has been re-uploaded, not filmed.

The tool runs entirely against YouTube's metadata; nothing about the source video is uploaded to Amnesty servers. The user provides a URL, the tool queries YouTube, the user receives the metadata. Operational simplicity is the point: the tool does one thing, does it reliably, and pairs cleanly with reverse- image search elsewhere in the workflow.

When to use it

  • A YouTube clip is being claimed as fresh footage from a current event, and the fact-checker needs to verify the upload timestamp against the claimed event time.
  • A video has been re-uploaded under a misleading title or caption, and the thumbnail set needs reverse-image-searching to locate the original posting.
  • An archival pipeline (Auto Archiver, InVID-WeVerify's WACZ module) needs the precise upload timestamp recorded as evidence metadata before the source URL is preserved.
  • A workshop is teaching the Pillar 2 source-history pillar in practice and needs a tool that demonstrates metadata-driven verification without requiring file uploads or paid access.

Limitations

Limitations

  • YouTube-only. The tool does not work on Facebook video, TikTok, Twitter/X video, Telegram-hosted video, LINE-shared content, or any other platform; for non-YouTube video the user must reach for InVID-WeVerify's keyframe extraction or the platform's own metadata if available.
  • Dependent on YouTube's metadata API behaviour. If YouTube changes how upload timestamps are exposed or restricts thumbnail access, the tool's output changes accordingly.
  • Returns metadata, not verification; the upload time and thumbnails are inputs to a manual chain (reverse-image search, cross-platform comparison, claim-database lookup), not a finished verdict.
  • English-language interface only; the tool itself is language-agnostic on input because YouTube URLs are not language-bound.

Privacy and threat model

The tool's privacy posture is unusually clean: no source file is uploaded, only a YouTube URL is submitted. The query goes to YouTube via Amnesty's web tool; no analysis of the video itself takes place on Citizen Evidence Lab servers. For surveillance- environment work this is the operationally safe path; using the YouTube Data Viewer does not expose the source video to a third-party analysis pipeline because no such analysis happens.

The downstream reverse-image-search step that uses the returned thumbnails does involve querying third-party search engines (Google, Bing, Yandex, Baidu, TinEye), which carries the usual search-engine query-logging considerations. The thumbnails themselves are public YouTube assets so the privacy delta is minimal compared to a normal browser session.

Country and platform applicability

  • Indonesia: language-agnostic; useful for verifying YouTube re-uploads of political clips during election cycles. Pairs with the MAFINDO/CekFakta workflow at Kalimasada.
  • Laos: language-agnostic; YouTube content from or about Laos is a marginal use case (YouTube reach in Laos is limited compared to Facebook), but the tool works on any YouTube URL regardless of upload origin.
  • Malaysia: language-agnostic; useful for Sebenarnya.my and Bernama MyCheck verification of Malaysia- facing YouTube content.
  • Philippines: language-agnostic; documented heavy use across YouTube-uploaded political clips in the #FactsFirstPH workflow context, particularly for re-upload detection.
  • Sri Lanka: language-agnostic; useful for Watchdog and Hashtag Generation verification of YouTube clips during election cycles.
  • Thailand: language-agnostic; pairs with Thai PBS and SONP verification work on YouTube-uploaded political content. LINE is the dominant Thai messaging platform but YouTube is still the primary long-form video host.

Platform applicability: YouTube only. The tool's narrow scope is its design.

How to access

Free web tool at citizenevidence.amnestyusa.org or the Citizen Evidence Lab project page. No account creation, no install, no paid tier. The tool has been online and stable since the early 2010s as part of the Citizen Evidence Lab's broader verification-resources hub.

Cost (current as of 2026-05)

Free. No paid tier exists; the tool is part of Amnesty International's public-good verification infrastructure.

Quickstart

  1. Copy the YouTube URL of the video you want to verify.
  2. Paste it into the YouTube Data Viewer at citizenevidence.amnestyusa.org.
  3. Read the returned upload timestamp: the precise time the video was uploaded to YouTube, not the rounded date YouTube displays publicly.
  4. Compare the timestamp against the claimed event time; an upload that predates the claimed event is one strong non-detector signal that the clip is misattributed.
  5. Take each returned thumbnail and reverse-image-search it (InVID-WeVerify's reverse-image module is the standard pairing) to locate earlier instances of the video.
  6. Document upload time and reverse-image-search hits as evidence in the fact-check workflow.

In the toolkit's workflow

Source-history pillar non-detector tool per Architectural Anchor 1. Sits in 2A.2 as the YouTube-specific source-history alternative alongside the cross-cell primary InVID-WeVerify (broader reverse-image and keyframe pipeline), the AI-geolocation alternative GeoSpy, and the archiving infrastructure Auto Archiver.

Standard combinations:

  • With InVID-WeVerify at 2A.2 / 1B.1 – Citizen Evidence Lab YDV returns the upload timestamp and thumbnails; InVID's reverse-image module then searches the thumbnails across multiple engines in one step.
  • With Auto Archiver at 2A.2 to preserve the source URL plus the verified upload timestamp as a single evidence record before the YouTube clip can be removed or edited.
  • With GeoSpy at 2A.2 when a thumbnail or extracted frame from the YouTube clip needs a geographic-hypothesis starting point.
  • With Meedan Check at 2B.1 when the verified YouTube clip enters a tipline workflow and the upload timestamp is part of the fact-check response back to the community.

Decision-tree references: T2 video triage routes YouTube-sourced video through Citizen Evidence Lab YDV at T2.2 (platform preservation) and T2.4 (keyframe extraction). T1 image triage routes any extracted thumbnail through reverse-image search via the standard InVID pairing.

This card carries no detector signal class: the tool produces metadata (upload time, thumbnails), not a synthetic-content verdict. Per Anchor 1 the metadata feeds the source-history pillar; per Anchor 2 an upload-time mismatch by itself is one non-detector signal that combines with reverse-image hits and archived-version checks to support a publishable claim.

Override notes

Override notes

  • Non-detector declaration: the tool returns YouTube metadata, not a content-classification verdict. It serves the source-history pillar; the analytical work (cross-checking timestamps and reverse-image-searching thumbnails) is the user's, not the tool's.

Sources