What’s Out There
← Back to swiping
Social mediaBy applicationAcademic

Lumen Database

Berkman Klein Center for Internet and Society, Harvard University

Millions of takedown demands sent to Google, X, and others, including who wanted what removed.

Lumen collects copies of legal demands to remove online content, mostly DMCA copyright notices but also defamation, trademark, and court orders, submitted by companies such as Google. Each notice lists sender, recipient, claimed legal ground, and the targeted URLs. The public can search; full notices in bulk need researcher credentials.

Access

Cost
By applicationFree search; the public can view one full notice per email per day, and researchers apply for API credentials.
License
Provider's own termsLumen terms; API terms for researchers
Sign-up
Application required
How you get it
APIHTML, JSON

What's in it

One row is
One row per takedown notice, with listed URLs
Covers
Global, mostly United States
Years
2002-present
Updated
Daily
Size
Not stated
senderprincipalrecipientnotice typedatejurisdictiontargeted URLsworks claimed

Quality

4 of 5

The only large public archive of takedown demands; coverage depends on which companies choose to share their notices.

  • Presence in Lumen says nothing about whether the content was removed
  • Credentials are granted at Lumen's discretion

Connects to

Questions it could answer

  • Which senders file the most fraudulent-looking DMCA notices?
  • How have defamation-based removal requests grown relative to copyright notices?

Similar data sets

  • DSA Transparency Database

    Billions of content moderation decisions by online platforms, filed almost in real time under EU law.

    Social mediaEuropean Commission (DG CONNECT)Free

  • Luminate Music Data (CONNECT)

    The sales, streaming, and airplay counts that have powered the Billboard charts for 30-plus years.

    Audio & musicLuminate (formerly Nielsen Music / MRC Data)Paid

  • Labeled Faces in the Wild (LFW)

    13,000 news photos of 5,700 public figures, gathered without consent, that set face-recognition benchmarks.

    Images & videoUniversity of Massachusetts AmherstFree

  • Google Transparency Report: Government Requests to Remove Content

    Every six months, Google counts which governments asked it to take content down, from where, and why.

    Social mediaGoogleFree

  • Meta Content Library

    Public Facebook, Instagram, Threads, and WhatsApp Channels posts, for vetted researchers only.

    Social mediaMeta PlatformsBy application

  • The Pile

    The 825 GB open training set that taught the field how LLM data is assembled, and got sued over it.

    Language & textEleutherAIFree