This week: Amnesty International's forensic account of how coordinated "foreign agent" disinformation under Indonesia's Prabowo government moved from social media accounts to a human rights defender burned by acid in Jakarta. Then a look at Armenia's parliamentary elections, held today, and what it means that Russia's interference playbook was documented in writing before the polls opened. And the read past the Lancet's fabricated citation study: the failure mode that automated checkers will never catch is already the dominant one.
The thread connecting all three this week is the distance between documentation and prevention. Indonesia shows what happens when the forensic record is complete: campaigns traced, arrests made, platforms slow to act regardless. Armenia shows what happens when interference is named before it occurs: a documented prediction, now waiting for the outcome to test it. The Lancet read asks what happens when the detection tools get good enough to catch the obvious failure, and the subtler one moves into the space they vacate. All three are about what the record can establish, and what that establishment does not automatically prevent.
Indonesia's military ran coordinated "foreign agent" campaigns, and an activist ended up in hospital
Andrie Yunus spent months being labelled online as a foreign agent before someone burned him with acid on a Jakarta street in March 2026. He is a human rights investigator at KontraS who had led protests against revisions to Indonesia's military law. The online campaigns targeting him began in March 2025, when an account affiliated with President Prabowo Subianto's Gerindra party uploaded a video labelling him and other activists as foreign agents trying to weaken Indonesia's armed forces. Amnesty International's metadata analysis found the video was first uploaded by three Gerindra party office accounts before being amplified by 31 accounts affiliated with 27 separate military units across Instagram, Facebook, X, and YouTube. The investigation into the acid attack led to the arrest of four officers from the Indonesian Strategic Intelligence Agency.
Amnesty's May 2026 report, "Building Up Imaginary Enemies," documents four distinct disinformation campaigns across the first 18 months of the Prabowo government. Three involved social media accounts presenting themselves as military units; the fourth targeted the Centre of Economic and Law Studies. The campaigns shared a structural pattern: highlight international partnerships or foreign grants, present them as evidence of a hidden agenda, amplify through coordinated identical posting across platforms, then wait for the chilling effect or the physical consequence. Tempo, one of Indonesia's most respected news organisations, received a severed pig's head at its newsroom after sustained campaigns accusing it of being controlled by foreign donors.
Meta told Reuters it had disrupted multiple adversarial networks targeting Indonesia and has removed over 200 such networks globally since 2017. TikTok pledged additional monitoring. X and YouTube did not respond to Amnesty.
The attribution chain from online campaign to physical attack has now been tested in an Indonesian court. The question for OSINT practice is whether Amnesty's metadata methodology is documented in enough detail to be reproduced by other investigators working on comparable cases.
The framework matters beyond Indonesia. The "foreign agent" narrative is a documented template: establish international funding links, cast them as evidence of hostile intent, coordinate amplification across multiple platforms using accounts that present as institutional rather than political. The Amnesty report is notable for what it establishes about the coordination layer, not just the content. Accounts presenting as military units posting simultaneously from different locations; identical captions appearing on unrelated accounts within short windows; content seeded from partisan accounts before amplification by officially-affiliated ones. These are the forensic signatures. An investigator working on comparable cases in another jurisdiction now has a documented precedent for what coordinated state-adjacent disinformation looks like when it reaches accountability level.
Today, 4 June, The Jakarta Post published a companion investigation into whether the August-September 2025 protests themselves were framed as foreign-backed by pro-Russia and pro-China actors. The study, a consortium of Indonesian media outlets and the social media monitoring platform Drone Emprit, found no evidence that Open Society Foundations funded or organised the protests. What it found instead was a campaign designed to make that claim appear credible. The disinformation-about-the-disinformation layer is now documented. Both pieces are linked below.
