An investigation by the Georgian fact-checking platform Myth Detector has identified 230 fake Facebook accounts with AI-generated profile pictures that appear to be operating as a coordinated network to amplify narratives supporting the ruling Georgian Dream party. According to the organization, the accounts demonstrated signs of coordinated inauthentic behavior (CIB), with their activity concentrated on boosting specific talking points from the party and its chairman, Irakli Kobakhidze.
The accounts shared similar photos, names, and behavioral patterns, the analysis found. Several also listed authentic workplaces and educational institutions in their profiles - a detail designed to make them appear credible to casual observers.
What triggered the network
The activity intensified around an August 12 briefing by Kobakhidze marking the 18th anniversary of the Russia-Georgia war. At that briefing, Kobakhidze focused on what he termed rising "Russophobia" in Georgia, and linked the 2008 war to the "deep state," according to Myth Detector. He cited rallies against Russian tourists and protests over the arrival of Russian citizens in Georgia as examples of the phenomenon.
Following the briefing, pro-government television channels Imedi and Rustavi 2, along with the Interpressnews agency, published informational cards on Facebook focused on the alleged stoking of anti-Russian sentiment and the role of the "deep state." Myth Detector examined the "likes" on these posts and found suspicious profiles with identical behavioral patterns that appeared to be operating in concert to artificially amplify support for those narratives.
What the analysis found
Myth Detector said the 230 accounts shared characteristics pointing to coordination, including AI-generated profile pictures, duplicate imagery, fabricated demographic diversity, real employers and schools listed in bios, and near-identical behavior:
- AI-generated profile pictures. According to the AI-detection tool Hive Moderation, all pictures from the 230 accounts have a greater than 90% probability of being AI-generated, reaching 100% in some cases. Visual inspection also found characteristic AI artifacts, such as unnaturally shaped hands and fingers.
- Duplicated photos and compositions. Some accounts used identical or highly similar imagery, including the exact same photographs with minor edits. In several instances, identical first and last names appeared on separate accounts.
- Artificially created demographic diversity. The network created the appearance of endorsement from people of varying ages, appearances, and lifestyles: profiles of young and elderly users, conservative and alternative-looking citizens, believers, family-oriented individuals, athletes, and vacationers.
- Real workplaces and educational institutions listed. Most accounts listed legitimate employers and schools, including Tbilservice Group, the Tbilisi Transport Company, Georgian insurance companies, commercial banks, microfinance organizations, state agencies, and municipal services.
- Identical behavioral patterns. Most profiles listed their move to Georgian cities as occurring between July 20 and July 24, 2026; some older accounts were also identified.
- Activity focused on identical talking points. Reactions from the suspicious accounts concentrated on posts promoting conspiracy statements by the ruling party chairman regarding the "Global War Party," "deep state plots," "opening a second front," and "Russophobia" narratives - pointing collectively to artificially inflated engagement.
Not the first such operation
Myth Detector emphasized this is not the first time AI-generated imagery has been detected in such operations. The organization observed similar AI-profile activity in April, when 75 fake pages actively reacted to posts by Russia-based channel Rustavi 2 concerning treatments for Duchenne muscular dystrophy. Of those 75 accounts, 47 used modified versions of the exact same AI-generated photo. Several of those accounts had previously participated in various smear campaigns.
The findings raise practical questions for professionals working in AI for PR & Communications and AI for Government, where detecting manipulated content has become a core operational requirement.
Why this matters for PR and communications professionals
This case shows a playbook that public-facing organizations in any country can expect to encounter: AI can mass-produce realistic - but fake - supporters, and coordinate them around a single message within hours. For communications and government professionals, verifying who is actually engaging with official statements is no longer a matter of looking at profile photos. Checking for coordinated patterns - duplicate images, repeated timestamps, identical storytelling points - needs to be part of standard monitoring if you don't want to mistake a bot swarm for public sentiment.
Your membership also unlocks:
