Tibet's story is under attack — digitally.
#TransnationalRepression x Coordinated Inauthentic Behaviour ( #disinformation ) are targeting #Tibetan voices & networks.
@tenam.bsky.social Call to Action is essential reading:
phayul.com/opinion-the-... #spamouflage
@globaltimescn.bsky.social now directly feeding on, & amplifying, #spamouflage bsky.app/profile/tibe...
Bottom line from the EEAS for #Tibetan #FIMI researchers:
#Tibet is a documented, sustained target of #AI-enhanced PRC state information operations. The EU's own foreign affairs service is now saying so publicly.
(& for Bluesky: #Spamouflage is on #Bluesky.)
THE GALAXY OF FIMI OPERATIONS IN 2025
The Galaxy of #FIMI Operations maps 3,000 core channels.
The report identifies distinct Chinese-attributed infrastructure: #Spamouflage, CGTN-linked #Falsos Amigos, #Paperwall, #HaiEnergy, alongside Russian-attributed channels.
[p.25, p.28-29]
#Spamouflage 'does not follow a single, consistent operational model.' It deploys account clusters mobilised for specific purposes: targeting #dissidents, responding to breaking news, advancing Chinese propaganda points. [p.29]
The #Bluesky example involves a video using a #dissident's identity to falsely accuse the Spanish government and the respected NGO @safeguarddefenders.bsky.social of corruption.
#Spamouflage uses AI to make such impersonations more convincing. (EEAS) [p.34] #transnationalrepression
Figure 20_Example of impersonation video by #Spamouflage on #Bluesky
The EEAS report includes an example: Figure 20 shows an AI-assisted #Spamouflage impersonation video posted to #Bluesky.
The new old: Spamouflage content creationThe IMS Spamouflage consists of newly created, repurposed,or hijacked53 social media accounts that post and amplifystate messaging across more than 50 platforms54, engagingalmost exclusively with other Spamouflage content. In2025, Spamouflage perfectioned a new attack pattern: theuse of AI-assisted impersonation videos. This buildson its previous creation of poorly edited pictures andis specifically employed to degrade dissidents anddiscredit opposition voices.Spamouflage appears to use AI to produce more content.Its accounts frequently share AI-generated cartoons builtaround similar narratives, allowing them to quickly generatelarge amounts of tailored material targeting specific actorsor issues.Even if overall engagement remains low, the repeatedposting of varied content creates the impression thatthese views are widely shared. For example, this techniquewas used to target both the EU and the US, with at least50 AI-generated cartoons portraying the EU as subordinateto the US.Spamouflage has also used AI to refine its impersonationtactics and vary its attacks. The IMS manipulated videosusing AI tools to imitate opposing voices. This marks a stepbeyond its earlier use of poorly edited images targetingindividuals.
EEAS: In 2025, #Spamouflage perfected a new attack pattern: the use of #AI-assisted impersonation videos. This builds on its previous creation of poorly edited pictures and is specifically employed to degrade dissidents and discredit opposition voices. #transnationalrepression
SPAMOUFLAGEThe Chinese-aligned IMS47 is only partially representedin the graph for illustrative purposes, as it comprises alarge number of inauthentic accounts that would otherwiseovercrowd the visualisation. The IMS is mostly activeon X, where its activity is organised around clustersof accounts. These clusters are characterised by theirown behavioural patterns, and are mobilised for specificpurposes, like responding to breaking news events, targetingdissidents or advancing Chinese propaganda points. Asa result, the IMS does not follow a single, consistentoperational model. One prominent pattern of Spamouflageinvolves a first cluster of seeder accounts impersonatingdissidents, which post a video or an image containing falseallegations. Then, a second cluster of amplifier accounts thatdisseminate the content by posting it as a reply to relevantentities, such as government bodies accounts. Finally, athird cluster of accounts posts, also as a reply to similarentities, a screenshot of the initial video and hashtags tocriticise and call for action against the dissident who allegedlypublished the content. A later section of the report examinesSpamouflage’s emerging use of AI-generated impersonationvideos.
#Spamouflage, the Chinese state-linked influence network, operates across more than *50 platforms.* #genAI
🧵 The EU's foreign affairs service just published its 4th annual report on foreign information manipulation. #FIMI
It identifies China's #Spamouflage network as active here on #Bluesky (大家好!) It touches on the targeting of #Tibetans. A thread on what it says:
Screenshot of a QT on Twitter: Jurre van Bergen @DrWhax China does this with Meiya Pico, now called SDIC Intelligence owned by a state investment vehicle. It's software has been battle tested in places like Tibet and against the Uyghurs. For background: https://tibetwatch.org/a-long-shadow-the-expansion-and-export-of-chinas-digital-repression-model-in-tibet/ https://therecord.media/chinese-firm-tied-to-uyghur-abuses-training-police-hacking-tibet China Now @ChinaNow24 The Government of the United Kingdom is purchasing Israeli surveillance software tested on Palestinians. The contracts reportedly include technologies from Cellebrite, BriefCam, and Corsight for phone hacking, facial recognition, and mass surveillance. In the West, nothing seems to rival importing surveillance tools tested in conflict zones—under the banner of “defending human rights.” Source: https://x.com/DrWhax/status/2030460883894510049
@turquoiseroof.org doesn't spend much time on X (unless we are tracking #spamouflage) but we appreciate this post over there from Amnesty's @jurrevanbergen.nl linking to our report (w/ @tibetwatch.org) in this context turquoiseroof.org/a-long-shado... | ZH summary turquoiseroof.org/download/%e8...
@Graphika.com attributes PRC #disinformation targeting the Sikyong elections to #Spamouflage bsky.app/profile/jamy...
#Graphika says Tumblr accounts it assessed with high confidence as part of the Chinese state-linked influence operation #Spamouflage ( #公安部912专项工作组) published AI-generated articles simultaneously in English and Tibetan targeting candidates in the CTA election.
Glass Onion: @graphika.com exposes the commercialised machinery of Chinese influence operations: PR firms, fake “foreign media” x #Spamouflage amplification. graphika.com/reports/glas...
Yao Zhang knew something was up when she started seeing fake sexually explicit images of her circulating online. The YouTuber, who often criticizes the Chinese government, is one of several Canadians targeted by a #spamouflage campaign led by the People's Rep. of China.
www.cbc.ca/news/canada/...
7/ An early test for China's future operations
The operation showed no signs of certain TTPs common in Chinese #IOs on HK-related issues, suggesting that it was merely testing the waters
Eg. Cross-posting via 50 Cent Army/ #Spamouflage
Amplification by major KOL/ #GrayMedia in HK
"A set of inauthentic accounts that post content. Lots and lots of content."
That's the Chinese state-linked operation #Spamouflage as described by Graphika founder Dr. John Kelly in our recent webinar on foreign information operations.
To watch the full video: graphika.com/reports/how-...
What - and who - is #Spamouflage?
Graphika Founder and Executive Chairman Dr. John Kelly discussed the Chinese state-linked actors in our recent webinar focusing on foreign information operations.
To watch the full video on demand: graphika.com/reports/how-...
6 posts. 4,800+ exposures. 📈
Graphika’s ATLAS platform uncovered how a pro-China trade narrative spread online — and who amplified it.
🎥 Watch the breakdown! Signup for a free demo here: graphika.com/get-started?...
#ThreatIntel #Disinformation #Spamouflage #OSINT
Fact-checking brings out the cyberstalkers, so CTV's answer is to stop fact-checking??? During an election, we need better information not cowardice.
youtube.com/watch?v=HdPi...
#CTV #cyberstalking #CdnPoli #ItsOurVote #Canada #ElectionsCanada #ElectoFacts #truth #transparency #Spamouflage
4/7
The campaign—described as "spamouflage" where content is mixed with everyday material—marks "the first known instance where a Spamouflage campaign used sexually explicit deepfake photos to target an individual in Canada," Global Affairs Canada reported.
#Deepfakes #Cyberthreats #Spamouflage
"The state-linked #Spamouflage operation ... has repeatedly targeted the Spain-based non-profit Safeguard Defenders and in January posed as the organization to spread online calls for the Spanish government to be overthrown in response to deadly floods in Valencia."
graphika.com/reports/chin...
Chinese influence operation urged Spaniards to 'overthrow the government' www.euronews.com/my-europe/20... #Spain #SpainFloods #ValenciaFloods #Graphika #Spamouflage #China #InfluenceOperation
Accounts associated with the #Spamouflage op, posed as the @safeguarddefenders.bsky.social on social media platforms including Facebook, X, BlueSky and TikTok from November to January, according to this @graphika.com report:
Politico: #China -linked influence operation tried to overthrow Spain’s govt, report says
Fake accounts associated w/the #Spamouflage operation fanned discontent against establishment figures following floods in the Valencia region that killed more than 200 people. www.politico.eu/article/chin...
'An operation dubbed #Spamouflage posed this month as human rights group Safeguard Defenders to spread online calls for the Spanish government to be toppled in response to the catastrophic floods in October.'
#Spanje #Valencia www.reuters.com/world/chines...
A collaborative report by @doublethinklab.bsky.social and VOA on Chinese information operations in cyberspace against the U.S. presidential election.
#spamouflage #magaflage #USElection2024
Ingérence électorale : la présidentielle américaine dans le viseur de Pékin. Deux études, de @isdglobal.bsky.social et Graphika, soulignent la montée en puissance de l'opération d'influence #Spamouflage, orchestrée par le régime chinois 👇
infox.fr/ingerence-el...
〝The batch accounts — the 10 new accounts — are not very active. Each account has roughly 100 posts or reposts over the last two years. The batch accounts were inactive for one year but were awoken after the first Trump assassination attempt.〞
#China #USA #elections #infops #Spamouflage #MAGAflage
〝The accounts Graphika identified bore many hallmarks of Spamouflage activity, including coordinated posting and sharing content that has previously been linked to the operation. Some accounts occasionally slipped up and posted in Chinese.〞
#TikTok #platforms #Spamouflage #PRC #USA #infops #China
