Skip to content

Media Cloud

FreeInstitutional

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

TL;DR

Free, open-source academic platform indexing 60,000-plus news sources and roughly two billion stories across 20-plus languages, with attention and network-analysis tools for tracking how narratives move through the news ecosystem. The 2C.1 free academic alternative for institutional partners and research labs that cannot fund Information Tracer.

What it does

Media Cloud is a research-grade open platform built and maintained across UMass Amherst, Northeastern, and Harvard's Berkman Klein Center. It crawls 60,000-plus news sources, indexes the full text of approximately two billion stories, and provides tools for attention analysis (how much coverage a claim or actor received, when, in which outlets), network analysis (which sources cite or link to which others), and cross-language clustering. The platform is the most widely used academic infrastructure for studying narrative flow through news media at scale; available evidence records 167 academic papers using Media Cloud data.

For a 2C.1 monitoring operation, the value is structural: when Information Tracer is out of budget, Media Cloud is the nearest free alternative for tracking cross-source narrative spread. The trade-off sits at the data layer rather than at the methodology layer; Media Cloud's news-source corpus is structurally thinner for SEA languages than for English-language coverage, which the toolkit flags as the cell-level honest-gap on this entry.

When to use it

  • An academic partner or research lab is producing a study on how a claim moved through the news ecosystem and needs a queryable corpus of multi-year, multi-source full-text coverage.
  • An institutional fact-check coalition wants to compare how a cluster of false claims was covered across earned media in different countries and languages, as the news-side complement to social-platform spread analysis.
  • A regional civil-society partner is preparing a published investigation on a long-tail narrative and needs methodologically rigorous attention-curve evidence rather than per-message tipline data.
  • An institutional research budget cannot fund Information Tracer but the operation has the analyst capacity to build the news-side monitoring layer on Media Cloud's open infrastructure.

Limitations

Limitations

  • Limited SEA-language coverage in the news-source corpus. The platform indexes 20-plus languages but coverage of Thai, Bahasa Indonesia, Filipino/Tagalog, Sinhala, and Lao is structurally thinner than for English. For SEA-native monitoring, Sinar Project iMAP is the SEA-specific civil-society alternative; Media Cloud is best for cross- language and cross-region work where the SEA-side gap is acknowledged rather than concealed.
  • News-source-corpus focus rather than social-platform focus. Media Cloud surfaces what news outlets published; for what moved on Facebook, X, TikTok, or Telegram, the platform is not the right entry point; cross-link to Information Tracer (paid mid-tier cross-platform tracer) or to 1C.1 CIB tools (CIB Mango Tree, CooRTweet (with CooRnet predecessor)).
  • Quality of analysis depends on the data pipeline the analyst builds around the corpus. Media Cloud is academic-grade infrastructure rather than a no-code product; an operation without research-side analyst time will not get the same output quality from it.
  • Output is news-attention behavioural signal, not a verdict on claim truth. Per Anchor 2 the attention curve combines with a claim-extraction or fact-check signal from 2B before any publishable assertion about the underlying claim.

Privacy and threat model

Media Cloud queries run against a public-news corpus on academic infrastructure; the privacy concentration sits with the analyst's own query record on the platform rather than with content uploads that originated from a source. For institutional users the standard academic data-handling regime applies.

The threat model that matters here is editorial rather than data-flow. Attention-curve findings about how news outlets covered a claim in a specific country are research output that cited publication shapes; Indonesia's UU ITE/PDP, Malaysia's CMA Section 233 and 3R speech laws, and Thailand's lèse-majesté constraints each carry implications for what cited findings can be republished locally without secondary legal risk. The data path itself is open and academic; the cited use of the data is where local legal regimes come in.

Source-protection override (S2 — state-linked or legally sensitive)

Media Cloud attention curves on coverage of state actors or state-linked claims become published evidence whose circulation is subject to local speech laws. Mitigation steps:

  1. Consult Digital Safety, legal counsel, or the editor before publishing attention-curve findings on coverage of monarchy, military, police, ruling party, or named state officials in Thailand, Malaysia, Indonesia, Sri Lanka, or Laos.
  2. Treat the attention curve as a research artefact first; reproducing it in a local-language fact-check carries the legal exposure of the underlying coverage it visualises.
  3. Use the country-page operational guidance to decide which findings can be published locally versus held for cross- border or institutional partner channels.

Country and platform applicability

  • Indonesia: SEA-language coverage is thinner than English coverage; Bahasa Indonesia news-source coverage is partial. Useful for cross-language work pairing English-language coverage of Indonesian events with selected Bahasa sources.
  • Laos: structurally thin Lao-language news-source coverage. Honest gap: Lao operations needing 2C.1 monitoring will generally see thinner output from Media Cloud than from SEA-specific alternatives like Sinar Project iMAP.
  • Malaysia: partial SEA-language coverage; Malay and English Malaysian sources are partially indexed. Pairs with Sinar Project iMAP for the network-monitoring layer Media Cloud does not address.
  • Philippines: partial coverage; English-language Philippine sources are better-indexed than Filipino-language sources. Useful for #FactsFirstPH research-layer work pairing English Philippine media coverage with social-platform spread analysis.
  • Sri Lanka: partial coverage; Sinhala-language coverage is structurally thin per the broader Sinhala low-resource-language pattern (regional research). Tamil coverage is also thin on the Sri Lankan side specifically.
  • Thailand: partial coverage; English-language Thai sources are better-indexed than Thai-language sources.

Platform applicability: news-source corpus, not social-platform corpus. For news-side attention analysis on Facebook / Facebook Groups / TikTok / Telegram, this is not the right tool; Information Tracer or the 1C.1 CIB stack is.

How to access

Free registration at mediacloud.org. The platform has a web interface for query work and a documented API for programmatic extraction. Account creation requires institutional or research context; the academic origin of the platform shapes the access profile, but registration is open and not gated to a closed researcher network.

Cost (current as of 2026-05)

Free. Open-source. The platform is funded through the academic consortium of UMass Amherst, Northeastern, and Harvard's Berkman Klein Center plus grant support. There is no paid tier for ordinary use. The structural cost is analyst time to build a query pipeline against the corpus.

Quickstart

  1. Register a free account at mediacloud.org.
  2. Define the query: claim or narrative keywords, time window, country / language filter where supported.
  3. Run an attention-analysis query to surface the volume curve of coverage across the corpus.
  4. Drill into the source-list output: which outlets carried the coverage, what proportion was original reporting vs syndication.
  5. Run a network-analysis query to surface how the sources cite or link to each other on the topic.
  6. Export the data via the API for further analysis in Gephi (network visualisation) or for combination with social-platform spread output from Information Tracer.
  7. Document the methodology in the published investigation; cite the Media Cloud platform per its standard citation guidance.

In the toolkit's workflow

Behaviour-pillar non-detector tool per Architectural Anchor 1. Sits in 2C.1 Automated claim monitoring as the free-academic civil- society alternative alongside Information Tracer (paid mid- tier cross-platform tracer), Sinar Project iMAP (free open-source SEA-specific network-monitoring), and FactFlow AI (paid Telegram-specific alternative).

Standard combinations:

  • With Information Tracer at 2C.1 – Media Cloud surfaces news-side attention curves, Information Tracer surfaces social-platform spread; together they triangulate operations that move between earned media and social amplification.
  • With Sinar Project iMAP at 2C.1 for SEA-specific network-monitoring complementing Media Cloud's news-side coverage, particularly where the Lao or Malaysian information environment is the focus and the Media Cloud SEA-language gap is binding.
  • With Gephi at 2C.2 for publication-grade network visualisation of Media Cloud's source-network outputs.
  • With CIB Mango Tree and CooRTweet at 1C.1 for the social-platform CIB layer that Media Cloud does not address.
  • With ABCDE Framework at 2C.3 for codifying news-side attention findings under the Distribution axis.

Decision-tree references: T2 video triage names Media Cloud at T2.13 (cross-platform identical reuse) for the news-side attention layer; T5 escalation routes through Media Cloud when the case is multi-month or multi- country in scope and the coordinated-operation branch fires.

This card carries no detector signal class: Media Cloud produces a non-detector news-attention behavioural-pattern signal. Per Anchor 2 that signal combines with claim-extraction (2B.2), source-history (1B.1), or social-platform CIB (1C.1) to support a publishable institutional finding.

Override notes

Override notes

  • Limited SEA-language coverage disclaimer (b2-partial-pass-disclaimer-limited-SEA): the Limitations admonition records the SEA-language coverage gap concretely. The Country and platform applicability section names the thinness for each focus country. The recommendation is Sinar Project iMAP for SEA-native monitoring where the Media Cloud gap is binding.

Sources

  • Media Cloud. Media Cloud — open-source media analysis platform. UMass Amherst / Northeastern / Harvard Berkman Klein, 2024. mediacloud.org.
  • Media Cloud. Media Cloud Search. search.mediacloud.org.