Detect Deepfakesby Resemble AI
Deepfake case study · Image

Myth Detector: 230 AI-Generated Accounts Promoted…

An investigation by Myth Detector identified a network of 230 AI-generated Facebook accounts coordinating to promote narratives for the Georgian Dream party

Incident date
Aug 2026
Target
Georgian Dream party
Updated Aug 15, 2026 · 1 min read

An investigation by the Myth Detector platform has identified a network of 230 fake Facebook accounts engaged in coordinated inauthentic behavior to support the ruling Georgian Dream party in Georgia. These accounts utilized AI-generated imagery and synchronized activity to artificially amplify political messaging, particularly following a briefing by party chairman Irakli Kobakhidze on August 12, 2026.

What happened

The operation involved the deployment of 230 accounts that exhibited clear signs of synthetic creation and coordinated management. According to the investigation, the profile pictures for all these accounts were identified by the tool Hive Moderation as having a greater than 90% probability of being AI-generated, with some reaching 100% certainty. Visual inspections further confirmed the presence of AI artifacts, such as unnaturally shaped hands and fingers.

To establish credibility, the network populated profiles with real workplaces and educational institutions, including state agencies, commercial banks, and municipal services. The accounts created an illusion of diverse demographic support by varying the age and lifestyle appearance of the profiles. However, researchers noted suspicious behavioral patterns, such as multiple accounts claiming to have moved to Georgian cities within the same four-day window in July 2026.

The accounts operated in concert to boost engagement on posts from pro-government media outlets like Imedi and Rustavi 2. Their activity focused on specific talking points presented by Irakli Kobakhidze, including claims regarding a “deep state,” the “Global War Party,” and accusations of “Russophobia” in Georgia. This incident follows a similar pattern observed in April 2026, when 75 fake pages—many using modified versions of the same AI-generated photo—were used to react to posts regarding medical treatments. The current investigation concludes that the 230-account network functioned to artificially inflate support for specific government narratives through coordinated inauthentic behavior.

Sources