Cited by: @washingtonpost.com @nytimes.com @financialtimes.com @CNN.com @Wired.com @afpfactuel.bsky.social @LeMonde.fr @SZde.bsky.social @TheTimes.com @NBCnews.com @EFE.com
𝕏 / 🐦: antibot4navalny
The frequently-large Retweet, Like and View counts suggest inflated metrics—as 🟨 demonstrated by several engagement-for-hire bots that are also boosting metrics for 🟥 Matryoshka bots.
Which implies an actor with some money to burn rather than a grassroot activist.
The frequently-large Retweet, Like and View counts suggest inflated metrics—as 🟨 demonstrated by several engagement-for-hire bots that are also boosting metrics for 🟥 Matryoshka bots.
Which implies an actor with some money to burn rather than a grassroot activist.
Assuming one of the tweets has been lost somewhere in the pipeline, here are 46 of those 47 short tweets:
(originally posted in Armenian / machine-translated to English by me)
Assuming one of the tweets has been lost somewhere in the pipeline, here are 46 of those 47 short tweets:
(originally posted in Armenian / machine-translated to English by me)
As often happens with sloppy disinfo operators, a LLM prompt leaked out in one of the tweets, exposing both AI usage and Russian-language preference of the operator.
(original tweet in Russian / machine-translated to English by me)
As often happens with sloppy disinfo operators, a LLM prompt leaked out in one of the tweets, exposing both AI usage and Russian-language preference of the operator.
(original tweet in Russian / machine-translated to English by me)
They posted AI-processed snippets from @News.DW.com article on Russian state's increasingly aggressive disinfo efforts targeting Armenia parliamentary election in June 2026 ↘️.
They posted AI-processed snippets from @News.DW.com article on Russian state's increasingly aggressive disinfo efforts targeting Armenia parliamentary election in June 2026 ↘️.
As for underlying agenda, the fake video is a part of multi-month effort to target Pashinyan in anticipation of June-2026 elections in Armenia.
The DSA fine served as yet another pretext for attributing an intended fake narrative to a reputable org, in this case VIGINUM.
As for underlying agenda, the fake video is a part of multi-month effort to target Pashinyan in anticipation of June-2026 elections in Armenia.
The DSA fine served as yet another pretext for attributing an intended fake narrative to a reputable org, in this case VIGINUM.
Meanwhile, X/Twitter is effectively banned in Russia for over 4.5 years now, for non-compliance to Kremlin's censorship demands—which adds a new dimension to the hypocrisy of this fake video.
Meanwhile, X/Twitter is effectively banned in Russia for over 4.5 years now, for non-compliance to Kremlin's censorship demands—which adds a new dimension to the hypocrisy of this fake video.
To remind, the DSA fine has nothing to do with fake news on X.
In fact, the breached obligations quoted for the fine are limited to:
👮 deceptive Verification practice;
👮 non-transparent data on advertising;
👮 barring independent research of X/Twitter public data
To remind, the DSA fine has nothing to do with fake news on X.
In fact, the breached obligations quoted for the fine are limited to:
👮 deceptive Verification practice;
👮 non-transparent data on advertising;
👮 barring independent research of X/Twitter public data
As a further "proof", the fake video quotes (a non-existent) Trump's Jun-2025 exec order banning travel of those previously "involved in censorship".
The original exec order shown was aimed at "foreign terrorists" and other security threats: www.reuters.com/world/americ...
As a further "proof", the fake video quotes (a non-existent) Trump's Jun-2025 exec order banning travel of those previously "involved in censorship".
The original exec order shown was aimed at "foreign terrorists" and other security threats: www.reuters.com/world/americ...
A fake video from Matryoshka bots impersonates VIGINUM to allege that 65% of fake news (that "EC has fined X for") come from the very EU fact checking orgs.
A fake video from Matryoshka bots impersonates VIGINUM to allege that 65% of fake news (that "EC has fined X for") come from the very EU fact checking orgs.
As for Iyry Sofinn, he is a head of us-media[.]ru, a Krasnogorsk-based digital communication boutique.
He is also behind IQAD[.]ru (SEO agency) and SeoWizard[.]ru (a SEO service).
It is unclear at this point whether or not he runs the LLM-powered bot network.
As for Iyry Sofinn, he is a head of us-media[.]ru, a Krasnogorsk-based digital communication boutique.
He is also behind IQAD[.]ru (SEO agency) and SeoWizard[.]ru (a SEO service).
It is unclear at this point whether or not he runs the LLM-powered bot network.
Speaking of Dialogonomics(IuriKDmit), it seems like Lisiipen10U(Iyry Sofinn) is behind its operation:
🧩 Sofinn's 1st-ever post was a reply to IuriKDmit
🧩 which is also the very 1st reply received by IuriKDmit
🧩 Sofinn replied just 4m after IuriKDmit's very 1st post
Speaking of Dialogonomics(IuriKDmit), it seems like Lisiipen10U(Iyry Sofinn) is behind its operation:
🧩 Sofinn's 1st-ever post was a reply to IuriKDmit
🧩 which is also the very 1st reply received by IuriKDmit
🧩 Sofinn replied just 4m after IuriKDmit's very 1st post
Each of pro-Dmitriev LLM bots post up to 1-2 tweets per minute at peak.
One of the telling features for a bot network is using one account for a series of replies before switching to another account for a new series.
Each of pro-Dmitriev LLM bots post up to 1-2 tweets per minute at peak.
One of the telling features for a bot network is using one account for a series of replies before switching to another account for a new series.
This is the 1st time since Apr-2024 when is Kremlin using LLM bots that are context-relevant: responding to the subject of tweet (rather than spamming w/unrelated content)
Previously such bots were targeting Russian-speaking users, occasionally misunderstanding the context:
This is the 1st time since Apr-2024 when is Kremlin using LLM bots that are context-relevant: responding to the subject of tweet (rather than spamming w/unrelated content)
Previously such bots were targeting Russian-speaking users, occasionally misunderstanding the context:
The bots praise the special envoy responding to tweets posted:
👏 by Dmitriev himself
👏 by Dialogonomics (IuriKDmit), Dmitriev's analytical alter ego?
👏 by other users mentioning Dmitriev
As often, using bots to praise oneself is a telling sign of lack of confidence.
The bots praise the special envoy responding to tweets posted:
👏 by Dmitriev himself
👏 by Dialogonomics (IuriKDmit), Dmitriev's analytical alter ego?
👏 by other users mentioning Dmitriev
As often, using bots to praise oneself is a telling sign of lack of confidence.
Just a month later, Dmitriev is blaming EU today for using bots to manipulate his poll results.
Just a month later, Dmitriev is blaming EU today for using bots to manipulate his poll results.
CREDITS:
🙏 my stellar team exposed the entire phenomenon; did all of the heavy lifting on finding 100s of accounts, identifying common patterns; collecting ample evidence
🙏 Tatsiana Ashurkevich helped to structure and package our findings to make them easier to understand
CREDITS:
🙏 my stellar team exposed the entire phenomenon; did all of the heavy lifting on finding 100s of accounts, identifying common patterns; collecting ample evidence
🙏 Tatsiana Ashurkevich helped to structure and package our findings to make them easier to understand
X policy enforcmnt issues are similar to the major anon MAGA influencers:
• serial impersonation, despite blueticks
• stolen accounts helping w/longevity
• engment farming as the prevailing behavior of these accts
• manipulated growth (IFBAP, follow trains and other)
X policy enforcmnt issues are similar to the major anon MAGA influencers:
• serial impersonation, despite blueticks
• stolen accounts helping w/longevity
• engment farming as the prevailing behavior of these accts
• manipulated growth (IFBAP, follow trains and other)
We have no evidence yet to dis/prove either of possible nature / motivations of the operator(s):
🤔 domestic propaganda outsourced to cheap-labor regions
🤔 influence operation of a foreign actor
🤔 individual creators driven by the X Revenue Sharing
🤔 romance / PB scam
We have no evidence yet to dis/prove either of possible nature / motivations of the operator(s):
🤔 domestic propaganda outsourced to cheap-labor regions
🤔 influence operation of a foreign actor
🤔 individual creators driven by the X Revenue Sharing
🤔 romance / PB scam
From our limited observations, men in their 50-60s are those who most frequently respond to these "attractive MAGA female" accts.
If (½ YourAge + 7) social guideline could be applicable, ages targeted could be:
🏹 24..54 for Viet-Origins²
🏹 27..68 for French-Emigrants³
From our limited observations, men in their 50-60s are those who most frequently respond to these "attractive MAGA female" accts.
If (½ YourAge + 7) social guideline could be applicable, ages targeted could be:
🏹 24..54 for Viet-Origins²
🏹 27..68 for French-Emigrants³
Evidence for LLM / chatbot equivalent for the Vietnamese-origin² accts:
🧩 messages from LLM to operator
🧩 conversations contradicting to account's bio
🧩 Biden-era narratives assuming that Trump's 2nd term is yet to happen
🧩 placeholders that were posted unfilled
Evidence for LLM / chatbot equivalent for the Vietnamese-origin² accts:
🧩 messages from LLM to operator
🧩 conversations contradicting to account's bio
🧩 Biden-era narratives assuming that Trump's 2nd term is yet to happen
🧩 placeholders that were posted unfilled
Pointers to Viet (Cambodian) origin for the 2/ netwk:
🧩 "Hoa kỳ" in location (=US in Viet)
🧩 ChatGPT slips to Viet: a known issue for non-EN system default
🧩 loc transparency: Viet / Cambodia are the only non-US exceptions
🧩 2 accts repurposed from Viet clearly-bots
Pointers to Viet (Cambodian) origin for the 2/ netwk:
🧩 "Hoa kỳ" in location (=US in Viet)
🧩 ChatGPT slips to Viet: a known issue for non-EN system default
🧩 loc transparency: Viet / Cambodia are the only non-US exceptions
🧩 2 accts repurposed from Viet clearly-bots
Speaking of followers count stats:
🕸️ 15 of Potential-Fakes⁸ have over 5K followers each
🕸️ the largest Potential-Fake⁸ account nearly reaches 34K followers
🕸️ some of Vietnamese-origin² reach 1200
Speaking of followers count stats:
🕸️ 15 of Potential-Fakes⁸ have over 5K followers each
🕸️ the largest Potential-Fake⁸ account nearly reaches 34K followers
🕸️ some of Vietnamese-origin² reach 1200
One of the Potential-Fakes⁸ was also exposed by @Cen4InfoRes.bsky.social in mid-2024, which we only realized after discovering it on our own.
Originally @/Luna_2k24, it eventually became @/speaks_patriot with 30K followers, got suspended just a few days ago.
x.com/BenDoBrown/s...
One of the Potential-Fakes⁸ was also exposed by @Cen4InfoRes.bsky.social in mid-2024, which we only realized after discovering it on our own.
Originally @/Luna_2k24, it eventually became @/speaks_patriot with 30K followers, got suspended just a few days ago.
x.com/BenDoBrown/s...
Date of "becoming MAGA" among Potential-Fake⁸ accounts ranges between May 2024 and Oct 2025.
Date of "becoming MAGA" among Potential-Fake⁸ accounts ranges between May 2024 and Oct 2025.