WRLD

Search

#SyntheticIdentity
Krzysztof Graj Oct 16

𝗜𝗿𝗮𝗻’𝘀 𝗳𝗮𝗸𝗲 𝗜𝘀𝗿𝗮𝗲𝗹𝗶𝘀 – 𝗵𝗼𝘄 𝗮 𝗳𝗼𝗿𝗲𝗶𝗴𝗻 𝗶𝗻𝗳𝗹𝘂𝗲𝗻𝗰𝗲 𝗻𝗲𝘁𝘄𝗼𝗿𝗸 𝗲𝗻𝘁𝗲𝗿𝗲𝗱 𝗜𝘀𝗿𝗮𝗲𝗹’𝘀 𝗲𝗹𝗲𝗰𝘁𝗶𝗼𝗻 𝗱𝗲𝗯𝗮𝘁𝗲 Ahead of Israel’s 27 October election, hundreds of social-media accounts have been presenting themselves as ordinary Israelis while injecting political messages into genuine domestic conversations. They appear as women, reservists, secular voters, religious activists and frustrated citizens. Some attack Benjamin Netanyahu, others have previously promoted narratives associated with the Israeli right, while many focus less on parties than on a broader message: Israel is divided, its institutions cannot be trusted and its political system is approaching crisis. A detailed investigation by the OSINT company Golden Owl has mapped this activity as a substantial Iran-linked influence network. Its October analysis identified at least 694 accounts, with 445 attributed at high or medium-high confidence to the operation. Between 11 August and 11 October, the network generated at least 4.7 million views and more than 160,000 interactions. The researchers found no evidence that these accounts were simply a group of independent Iranian commentators expressing political opinions. Their assessment rests instead on coordinated behaviour, false identities, account histories, copied content, language patterns and platform metadata linking significant parts of the infrastructure to Iran. 𝗙𝗿𝗼𝗺 𝟯𝟱 𝗽𝗲𝗿𝘀𝗼𝗻𝗮𝘀 𝘁𝗼 𝗵𝘂𝗻𝗱𝗿𝗲𝗱𝘀 𝗼𝗳 𝗳𝗮𝗸𝗲 𝗮𝗰𝗰𝗼𝘂𝗻𝘁𝘀 The investigation developed in two stages. Golden Owl’s first report, published in September, identified an operational core of 35 accounts: 32 personas presenting themselves as Israelis and three Hebrew-language media-style channels. During the 30-day period measured in that study, active accounts generated 947,322 impressions and more than 9,000 interactions. The October investigation expanded the dataset dramatically. Researchers analysed more than 560,000 data points, including tens of thousands of accounts, relationships between them and more than a quarter of a million posts. The wider network showed repeated patterns of identical texts, synchronised activity and accounts created in batches. The numbers therefore describe not a sudden explosion from 35 to hundreds of accounts, but the gradual exposure of a much larger infrastructure surrounding the original core. 𝗧𝗵𝗲 𝗳𝗮𝗸𝗲 𝗜𝘀𝗿𝗮𝗲𝗹𝗶𝘀 Some of the strongest evidence comes from the histories of individual personas. One account had spent years posting exclusively in Persian under the handle “mardeankabooti” before reappearing in August 2026 as an apparently Israeli user called “Alon Strauss”, complete with an Israeli flag and Hebrew political messaging. Other personas presented themselves as Israeli professionals, activists or ordinary citizens while X transparency information showed connections through the Iranian version of the Android application or locations inconsistent with their claimed identities. The account “Yaara Shalev” presented itself as an Israeli woman concerned with democracy and peace. Investigators found that photographs used to construct the identity were inconsistent, including borrowed and apparently AI-generated imagery. A small Hebrew-language channel called KigVoice, previously identified in Israeli government reporting as part of an Iranian influence environment, repeatedly published material that later appeared on the much more visible Yaara account. In one sample, 14 of 40 KigVoice posts subsequently resurfaced as captions or opening lines in Yaara’s content. Another persona, “Miriam Kozak”, had built an audience of more than 18,000 followers and had already been exposed as fictitious by Israeli researchers in 2024. Its political messaging changed substantially over time, moving between narratives associated with different sides of Israeli politics. That inconsistency is important: the apparent purpose was not necessarily to build a stable ideological following, but to enter whichever political argument offered the greatest opportunity to intensify division. 𝗔𝗜, 𝗰𝗼𝗽𝗶𝗲𝗱 𝗺𝗲𝘀𝘀𝗮𝗴𝗲𝘀 𝗮𝗻𝗱 𝗜𝗿𝗮𝗻𝗶𝗮𝗻 𝗳𝗶𝗻𝗴𝗲𝗿𝗽𝗿𝗶𝗻𝘁𝘀 The Iran attribution is based on more than political content. Researchers identified accounts using the Iranian version of the X Android application, repeated connections to previously identified Iranian influence assets, copied posts appearing across supposedly independent personas within minutes, and Hebrew containing unusual grammatical errors or Persian and Arabic keyboard artefacts. Generative AI also appears to have become part of the production process. Some accounts inadvertently published fragments resembling chatbot instructions, while other posts displayed patterns consistent with machine-generated Hebrew. AI-generated videos and synthetic profile imagery further reduced the cost of maintaining large numbers of apparently distinct personalities. However, none of these indicators was treated as proof on its own. Golden Owl explicitly excluded accounts where the evidence was insufficient and notes that VPNs, travel and platform location labels can create misleading signals. The attribution rests on multiple independent indicators converging across the network. 𝗧𝗵𝗲 𝗴𝗼𝗮𝗹 𝗶𝘀 𝗻𝗼𝘁 𝗻𝗲𝗰𝗲𝘀𝘀𝗮𝗿𝗶𝗹𝘆 𝘁𝗼 𝗲𝗹𝗲𝗰𝘁 𝗼𝗻𝗲 𝗽𝗮𝗿𝘁𝘆 The network repeatedly exploited issues that genuinely divide Israeli society: responsibility for the 7 October attacks, the war, the hostages, ultra-Orthodox military exemptions, economic pressure, emigration and confidence in state institutions. It also circulated misleading information about the election itself, including false voting dates, claims that voting had been postponed and posts suggesting online voting mechanisms that do not exist in Israel’s parliamentary election system. The pattern suggests a strategy broader than conventional electoral campaigning. Rather than consistently promoting one political faction, the network appears designed to weaken confidence in the government, military, democratic institutions and the electoral process itself. An influence operation does not need to persuade Israelis to support a particular candidate if it can convince different groups that defeat would prove fraud, betrayal or systemic failure. 𝗥𝗲𝗮𝗹 𝗜𝘀𝗿𝗮𝗲𝗹𝗶𝘀 𝗮𝘀 𝘂𝗻𝘄𝗶𝘁𝘁𝗶𝗻𝗴 𝗮𝗺𝗽𝗹𝗶𝗳𝗶𝗲𝗿𝘀 One of the network’s most effective characteristics is that it does not depend solely on its own followers. Fake personas enter comment sections and political discussions involving real Israeli journalists, politicians and ordinary users. Authentic Israelis then repost or interact with the material, sometimes carrying it far beyond the original account. Golden Owl does not classify those genuine users as participants in the operation. In most cases they appear to be unwitting amplifiers. But once a message passes from a fabricated Iranian-operated persona to an authentic Israeli account, its foreign origin becomes much harder for the next person to recognise. 𝗧𝗵𝗲 𝟰𝟬,𝟬𝟬𝟬-𝗯𝗼𝘁 𝗳𝗶𝗴𝘂𝗿𝗲 𝗯𝗲𝗹𝗼𝗻𝗴𝘀 𝘁𝗼 𝗮 𝗱𝗶𝗳𝗳𝗲𝗿𝗲𝗻𝘁 𝗽𝗲𝗿𝗶𝗼𝗱 A much larger figure has also appeared in recent reporting. In the first two days after the October 2023 Hamas attack, social-media intelligence company Cyabra identified more than 40,000 fake profiles participating in online discussions across major platforms. Their content generated hundreds of thousands of interactions and had a potential reach estimated at 531 million profiles. Those numbers are important historical context, but they should not be merged with the current Golden Owl investigation. There is no public evidence demonstrating that all of those 2023 accounts and the 2026 Iran-linked election network belonged to one continuous command structure. The 531 million figure is also an estimate of potential exposure, not proof that 531 million people saw or believed the content. 𝗪𝗵𝗼 𝗶𝘀 𝗮𝗰𝘁𝘂𝗮𝗹𝗹𝘆 𝗯𝗲𝗵𝗶𝗻𝗱 𝘁𝗵𝗲 𝗼𝗽𝗲𝗿𝗮𝘁𝗶𝗼𝗻? No confirmed individual operators have been publicly identified. Names such as Yaara Shalev, Miriam Kozak and Alon Strauss are personas, not verified identities of the people controlling the accounts. Golden Owl says its evidence supports an association with the Islamic Republic of Iran and with previously documented Iranian influence infrastructure, but it does not establish which Iranian institution, intelligence service, military organisation or contractor directs the network. That distinction is essential. Public OSINT supports attribution at the Iran-linked operational level. It does not currently justify naming individual Iranian officials as the people commanding this specific campaign. 𝗧𝗵𝗲 𝗱𝗲𝗲𝗽𝗳𝗮𝗸𝗲 𝗶𝘀 𝗻𝗼 𝗹𝗼𝗻𝗴𝗲𝗿 𝗼𝗻𝗹𝘆 𝘁𝗵𝗲 𝘃𝗶𝗱𝗲𝗼 The Israeli case illustrates a broader change in influence operations. The synthetic asset does not have to be a spectacular deepfake of a politician. It can be an apparently ordinary citizen who comments on politics, argues with neighbours, describes personal experiences and gradually becomes part of an authentic community. That can be more difficult to recognise because the propaganda does not arrive looking foreign. It arrives looking local. And when hundreds of fabricated citizens enter the same national conversation, they can manufacture the impression that political anger, despair or distrust is more spontaneous and more widely shared than it really is. 𝗦𝗼𝘂𝗿𝗰𝗲𝘀 Golden Owl — Exposing the Iranian Regime’s Influence Machinery Against Israeli Society https://web-dev-production-7a35.up.railway.app/resources/research/Exposing-the-Iranian-Regime-Influence-Machinery-Against-Israeli-Society Golden Owl Golden Owl — The Other Face of War: Iranian Regime–Linked Manipulation of Israel’s 2026 Election Debate https://web-dev-production-7a35.up.railway.app/resources/research/Iranian-Regime-Linked-Manipulation-of-Israel-2026-Election-Debate Golden Owl The Jerusalem Post — Iran-linked social-media accounts posing as Israelis to target election https://www.jpost.com/israel-election-2026/article-909863 The Jerusalem Post Ynet — Fake Israelis, real influence: Iran-linked network infiltrates Israel’s election debate https://www.ynetnews.com/tech-and-digital/article/syp119nqqzg ynetglobal Calcalist — Iranian influence network exposed ahead of Israel’s election https://www.calcalist.co.il/calcalistech/article/s1q0ifijzl calcalist Vesty / Ynet — Iranian fake-account network and October update https://www.vesty.co.il/main/article/rkwokjssgg vesty Israel Ministry for Diaspora Affairs — Iran’s active Hebrew-language influence campaign https://www.gov.il/BlobFolder/reports/israel-and-us-strikes-in-iran-daily-status_2_3_26/en/mashlat_Roaring_Lion_israel-and-us-strikes-in-iran-daily-status_2_3_26.pdf Government of Israel Israel Central Elections Committee — official 2026 election timetable https://www.gov.il/en/pages/time--table-26 Government of Israel Channel NewsAsia — historical Cyabra analysis of more than 40,000 fake profiles after 7 October 2023 https://www.channelnewsasia.com/singapore/how-fake-news-israel-hamas-stokes-outrage-hatred-and-potential-violence-southeast-asian-tiktok-5603221 𝗞𝗲𝘆𝘄𝗼𝗿𝗱𝘀 #DISINFORMED #Episode20 #Iran #Israel #IsraelElection2026 #IranianInfluence #IranianDisinformation #ElectionInterference #ForeignInterference #FIMI #Disinformation #InformationWarfare #InfluenceOperations #FakeAccounts #FakePersonas #SyntheticIdentity #ArtificialIntelligence #AI #AIPropaganda #SocialMediaManipulation #Astroturfing #NarrativeManipulation #PoliticalDisinformation #ElectionDisinformation #X #Twitter #Bots #BotNetwork #KigVoice #YaaraShalev #MiriamKozak #DigitalInfluence #InformationManipulation #ElectionIntegrity #GoldenOwl #OSINT #MediaManipulation #WRLD

web-dev-production-7a35.up.railway.app
Exposing the Iranian Regime’s Influence Machinery Against Israeli Society
701
Krzysztof Graj Oct 3

IRAN: MEK/NCRI – when AI impersonated a real activist in live conversations For years, synthetic media has largely meant fabricated images, cloned voices and manipulated video. A case uncovered by Anthropic in September 2026 suggests a more consequential development: artificial intelligence being used not merely to imitate how a person looks or sounds, but to behave as that person in an ongoing conversation. Anthropic says it dismantled a distributed influence operation targeting Iranians inside the country and abroad that it linked to the People’s Mojahedin Organization of Iran, or PMOI/MEK, and its political front, the National Council of Resistance of Iran. The operators used a shared AI-agent platform to support impersonation, audience profiling, coordinated social-media activity, synthetic personas and the laundering of organisational content through apparently independent accounts. At least four people involved in the operation worked for official NCRI media outlets, according to Anthropic. The company also found references to committee approval processes and MEK leadership, but said it could not independently determine the precise level of central control. Cloning a person from 8,400 Telegram posts The most striking part of the operation involved a real activist whose Telegram identity was copied. According to Anthropic, the operators instructed Claude in Persian that it was now that person, then provided approximately 8,400 of the activist’s Telegram posts so the system could learn and reproduce his writing style. The AI-assisted account was subsequently used to conduct live political conversations with the activist’s existing contacts. Anthropic says that, to its knowledge, the people receiving those messages did not know they were interacting with an account being operated with AI assistance. This is materially different from a conventional chatbot or a static deepfake. The system was not simply producing a forged statement attributed to someone else. It was using a large archive of that person’s authentic language as behavioural training material, then applying the resulting imitation interactively. In practical terms, the deception moved from synthetic content to synthetic identity. Anthropic did not publish the identity of the impersonated activist, nor details of his contacts, and the public report does not establish what specific information the conversations produced. What it does establish is that the impersonation was operational rather than experimental: the cloned account was used in real exchanges with real people. From social-media monitoring to psychographic dossiers The impersonation sat inside a much broader targeting system. Anthropic says the network scraped more than 500 social-media channels and categorised individuals by location, age, profession, political orientation and arrest history. It then analysed approximately 51,944 archived messages from conversations to construct detailed psychographic dossiers on dozens of people inside Iran. The system was therefore not limited to broadcasting propaganda. It was also designed to understand specific audiences and individuals. Profiles could be used to decide what kind of message was most likely to resonate with a particular person, which political identity they appeared closest to and how they should be approached. Anthropic describes this as a structured targeting funnel rather than ad hoc social-media activity. That distinction is important in the Iranian context. Political activists, dissidents and opposition contacts inside Iran can face arrest and severe punishment. The report does not show that the dossiers directly led to arrests or other harm, but collecting and organising information such as arrest history and political affiliation adds a surveillance dimension to what would otherwise be described as an influence operation. A shared AI system for an influence network Anthropic found that the actors used a shared Claude-based platform named “Viktor”, with separate workspaces and persistent memory. Those memory files contained approved sources, prohibited words, account-management instructions and methods intended to reduce the risk of detection. One operator loaded MEK founding doctrine into the system as what the report called “strategic base data”, allowing others to reuse it across their work. This persistent-memory structure meant the AI did not have to be briefed from scratch for every task. Once doctrine, stylistic rules and operational constraints had been stored, the system could repeatedly produce material consistent with the network’s objectives. Anthropic says the same general playbook appeared across different workspaces despite the operators not sharing obvious account infrastructure. The network also used automated pipelines to coordinate Instagram posting schedules and tailor messages to different audiences. Some accounts initially avoided mentioning the Mojahedin at all, allowing material aligned with MEK/NCRI narratives to appear as neutral or independent political commentary. Content was distributed through the organisation’s own media ecosystem as well as accounts presented as ordinary news or activist profiles. Anthropic assessed the campaign as Category Two on its Breakout Scale: active on multiple platforms with distribution through NCRI-linked media properties and amplifier accounts, but without evidence of broad authentic penetration. Synthetic citizens and disguised organisational media The same infrastructure generated AI avatars presented as ordinary Iranians, complete with Persian audio, to promote Maryam Rajavi’s political programme. Anthropic says the synthetic nature of these personas was intentionally concealed. Operators also rewrote and redistributed content originating from MEK-affiliated media while stripping identifying features or presenting it through apparently independent accounts. The network’s political messaging targeted the Iranian government but also rival opposition currents, including monarchists and supporters of the Pahlavi camp. Anthropic documented fabricated video material attacking a member of the Pahlavi family and the use of the slogan “Neither Shah Nor Sheikh”, consistent with the NCRI’s broader competition with other opposition movements. That rivalry provides relevant context. Reuters describes the MEK as one of Iran’s principal organised opposition movements in exile and the dominant force behind the NCRI, led by Maryam Rajavi. Its relationship with other opposition currents is often contentious, while the extent of its support inside Iran remains uncertain. What can be attributed – and what cannot Anthropic’s attribution is unusually specific but still needs to be stated carefully. The company says its investigation linked the operation to MEK/NCRI and established that at least four participants worked for official NCRI media organisations. It also found a common operational playbook, committee review structures and repeated references to MEK leadership. On that basis, Anthropic said central tasking was likely. It nevertheless stopped short of saying it had proven that NCRI’s senior leadership directly ordered or managed every component of the operation. That caveat matters. The strongest public evidence supports describing GTG-84006 as a MEK/NCRI-aligned influence operation involving personnel from official NCRI media, rather than asserting that every action was directly commanded by the organisation’s top leadership. The broader significance of the case lies elsewhere. Earlier generations of synthetic media attempted to fabricate evidence: a photograph, a voice recording or a video. Here, AI was used to maintain an identity over time, remember doctrine, analyse targets and speak directly to real people while appearing to be someone they already trusted. The deepfake was no longer merely an artefact. It had become a participant in the conversation. Sources Anthropic — Countering misuse of AI: September 2026 https://www.anthropic.com/threat-intelligence-report-september-2026 AI Misuse Case Library — GTG-84006 activist impersonation case https://www.misuseofai.com/en/cases/gtg-84006-activist-impersonation/ Reuters — Who makes up Iran’s fragmented opposition? https://www.reuters.com/business/media-telecom/who-makes-up-irans-fragmented-opposition-2025-06-18/ Reuters — Iran’s divided opposition senses its moment https://www.reuters.com/world/middle-east/irans-divided-opposition-senses-its-moment-activists-remain-wary-protests-2025-06-19/ Reuters — NCRI and Maryam Rajavi https://www.reuters.com/world/middle-east/dissident-leader-abroad-urges-iranians-bring-down-khamenei-2025-06-24/ Keywords #DISINFORMED #Episode12 #Iran #MEK #PMOI #NCRI #IranianOpposition #Disinformation #InfluenceOperations #InformationWarfare #ArtificialIntelligence #AI #GenerativeAI #ClaudeAI #Anthropic #AIDeception #AIImpersonation #IdentityImpersonation #SyntheticIdentity #InteractiveDeepfake #DigitalImpersonation #PersonaCloning #BehaviouralCloning #Telegram #PsychographicProfiling #TargetProfiling #AudienceTargeting #SocialMediaManipulation #Astroturfing #SyntheticPersonas #NarrativeLaundering #CoordinatedInfluence #PoliticalInfluence #DigitalInfluence #OppositionPolitics #IranPolitics #OSINT #WRLD

www.anthropic.com
Countering misuse of AI: September 2026 / Anthropic
1057