- Filed
- September 9, 2026, 00:15 UTC
- Actor
- Iranian state-aligned / IRGC-linked operators (APT42 / CALANQUE ION and related clusters; Google GTIG attribution)
- Target
- International online audiences, U.S. and Western political discourse, energy and conflict narrative spaces
- Confidence
- documented Primary documents, court records, official disclosures
- Techniques
- Synthetic mediaAstroturfingPlain folksCard stacking
- Channels
- large language models (Gemini and others)
- text-to-image generators
- social media personas
- specialist narrative framing
Summary
Google Threat Intelligence Group’s Q3 2026 AI Threat Tracker (released 8 September) details Iranian state-aligned operators prompting models such as Gemini to generate highly detailed image-generation instructions—studio lighting, camera angles, skin texture—for photorealistic fictitious online personas, and to adopt expert identities (energy-market analysts, psychological-warfare specialists) when crafting state-aligned narratives. This builds on prior Meta disruption of Iran-linked AI meme networks posing as U.S. activists.
Analysis
Primary sourcing is Google Threat Intelligence Group’s AI Threat Tracker report ‘From Prompting to Autonomy: The Evolution of Adversarial AI,’ published 8 September 2026 and corroborated by contemporaneous coverage in Cyber Magazine, Ynet, and other outlets summarizing GTIG findings. The report states that Iranian actors have moved beyond basic prompting: they instruct language models to produce highly detailed prompts for image generators specifying studio lighting, camera angles and realistic facial textures in order to create photorealistic fictitious online personas. Separately, operators direct models to adopt specialist identities such as energy-market analysts or psychological-warfare experts and to incorporate advanced persuasion techniques into narratives aligned with Tehran’s objectives.
This is a clear instance of synthetic-media generation combined with astroturfing. The photorealistic personas are designed to appear as authentic individuals rather than state-linked assets, enabling transfer of credibility (plain-folks presentation) to messaging that would otherwise be discounted as foreign propaganda. Selective framing of energy markets, conflict dynamics or psychological themes without transparent origin constitutes card-stacking. The technique set is continuous with earlier Meta-documented Iran-linked networks that used AI-generated memes and posed as everyday Americans on Facebook and Instagram to push anti-Republican, immigration and Israel-Palestine content while routing through U.S./Canadian proxies.
The activity qualifies as influence/propaganda operations because the core mechanism is concealment of origin and artificial construction of credible interlocutors rather than transparent advocacy. GTIG’s documentation of the prompting patterns and model-use logs provides forensic-grade evidence; platform takedowns supply behavioral corroboration. Scoring the methods—detailed synthetic persona generation, specialist identity adoption, selective narrative construction—keeps assessment neutral; identical AI-assisted persona and narrative packages appear across other state and commercial actors.
Literacy counter: treat newly appearing ‘expert’ or ‘activist’ accounts that rapidly produce high-quality, thematically consistent visual and textual content on sensitive geopolitical or energy topics as provisional. Inspect image metadata or generation artifacts where available, linguistic uniformity across posts, and account creation patterns. Prefer primary platform and threat-intelligence reports that publish prompting examples, asset counts and infrastructure indicators over secondary political framing. When specialist personas surface in coordinated clusters, cross-check claimed expertise against independent professional footprints before accepting the voice as organic. Demand transparent origin disclosure for any account claiming analytic or activist authority on contested international issues.