

The Authoritarian Collaboration Index (AuthCollab Index) is a first-of-its-kind, fully automated database that systematically detects, tracks, and categorises transnational collaboration events between authoritarian regimes, authoritarian-leaning governments, and authoritarian-leaning opposition parties globally. The Index combines the precision of political science typology with the power of cutting-edge natural language processing (NLP), large language models (LLMs), and high-frequency open-source monitoring to map the ecosystem of global authoritarian entanglement.
​
The AuthCollab Index contains data from January 2024 onwards, featuring daily automatic updates to provide near-real time tracking of authoritarian collaboration globally. It includes collaboration events between 148 actors, across seven collaboration categories, extracted from approximately 400 million articles, analyses, press releases, communiqués, conference proceedings, and investigative reporting in over 100 languages, using scalable automated workflows.
​
Each entry in the Index includes a concise AI-generated summary, an LLM-assigned type label, involved actors, source links, date, and a unique identifier. Events range from discrete actions like high-level meetings or media partnerships, to broader patterns such as ideological alignment or narrative coordination.
The Index is searchable by actor, collaboration type, and keyword. Collaboration events in the Index are assigned one of the following type labels:
​
financial_hard: cross-border financing with patronage, corruption, or electoral influence (e.g. party, communications, media, or campaign funding).
financial_soft: non-corrupt trade, investment, business, or infrastructure deals.
diplomatic_hard: diplomatic moves to defend, empower, or legitimize authoritarian actors (e.g. votes, vetoes, lobbying, endorsements, high-profile visits, image rehabilitation); collective legitimation of autocratic norms/delegitimation of democratic norms.
diplomatic_soft: symbolic relations (e.g. routine visits, solidarity gestures, congratulations) with no substantive support.
network: network, platform, or alliance building.
propaganda_hard: disinformation, information operations, or media campaigns supporting another authoritarian actor or their goals; media, news, or communications partnership and collaboration.
propaganda_soft: shared narratives or ideological alignment in public messaging.
military_hard: pro-autocracy military interventions; military training and joint exercises; surveillance and repression technology transfer.
military_soft: routine arms sales; technology transfers.
sharinglearning_hard: active authoritarian knowledge-sharing (e.g. political training programs; peer-to-peer support for autocratization; political advisor exchanges).
sharinglearning_soft: passive imitation of illiberal tactics, or passing laws similar to those of the other authoritarian actor without direct coordination.
trxpolice_hard: joint or tolerated repression of dissidents, transnational repression (e.g. forced repatriation, red notice abuse, intelligence sharing on dissidents); collaboration on ‘anti-terrorism’.
trxpolice_soft: police/legal cooperation not directly targeting dissent.
Read the full companion analysis 'Mapping the Global Autocratic Ecosystem' for the AuthCollab Index here:
Read the press release here:
Sources were evaluated using a custom credibility metric based on Media Bias/Fact Check, with preference given to high-quality outlets. However, in niche or underreported cases, events may rely on lower-credibility or authoritarian state media, so checking sources is advised. Despite extensive QA, the Index may contain errors: a small number of false positives may persist. Users are encouraged to flag suspect entries via email at info@actionfordemocracy.org. We consider this a living index, subject to refinement.
