Spamouflage: AI interference in 2026 Tibetan exile elections
Meta dismantled a Spamouflage cluster of 90 Facebook accounts and 13 Instagram profiles targeting the 2026 Tibetan exile elections. The network distributed AI-generated content to discredit candidates of the Central Tibetan Administration, with engagement remaining marginal.
The narrative claims that candidates for the presidency of the Tibetan Parliament-in-Exile are incompetent, corrupt or illegitimate. In April 2026 Meta dismantled a Spamouflage cluster of 90 Facebook accounts and 13 Instagram profiles spreading AI-generated content. Organic engagement remained marginal.
Executive summary
Meta dismantled a Spamouflage network of 103 accounts (90 Facebook, 13 Instagram) in 2025 attributed to China-linked actors targeting the 2026 elections of the Tibetan exile Administration. The operation used AI-generated content to discredit candidates, though measured engagement remained marginal.
What is observed
Meta published a report confirming the dismantling of a Spamouflage cluster. The network comprised 90 Facebook accounts and 13 Instagram profiles operating in coordinated manner. The distributed content specifically targeted candidates in the 2026 Tibetan exile Administration elections. Forensic analysis identified use of AI text-generation technology to produce negative campaign messages. Accounts displayed technical indicators of inauthenticity (anomalous engagement patterns, synchronized creation). Meta attributed the operation to Spamouflage, a group documented as having links to Chinese government entities. Reach metrics showed limited overall engagement before removal.
What this does not prove
Detection and network dismantling does not prove operational effectiveness on the final ballot. Observed low engagement does not confirm secondary intent or reduced operational capacity. Attribution to Spamouflage, based on technical similarities and known tactic patterns, does not constitute irrefutable proof of command or direct orders from Beijing. Using AI to generate content does not demonstrate overall infrastructure sophistication level. Absence of public evidence of measurable electoral impact does not refute longer-term influence objectives or normalization of such practices.
Confidence level
Meta possesses direct infrastructure data and technical artifacts from dismantlement, supporting medium confidence on operation facts. Attribution to Spamouflage relies on pattern analysis and the group's documented history, without revelation of command communications. Absence of primary Chinese government documents and marginal engagement characteristics limit certainty on the actual scope of intent.
Methodological limits
This brief relies on the analysis of publicly accessible content (OSINT). Attribution is based on converging technical and editorial indicators, without access to the internal communications of designated actors. Volume data reflects content captured by our 567-source pipeline and does not constitute an exhaustive census.
How to cite this investigation
DisInfo Monitor (2026), "Spamouflage: AI interference in 2026 Tibetan exile elections", independent publication, disinfo-monitor.com/en/narrative/spamouflage-campagne-elections-tibetaines-2026-echec-relatif-mor0hfh8, first detected May 4, 2026, last updated May 4, 2026, accessed May 19, 2026.