This is a research prototype. The data and analyses are preliminary and not yet validated — we'd welcome your .
A network of approximately 120 fake Facebook profiles, utilizing AI-generated profile images, was identified coordinating the amplification of political content for Bulgaria's 'There Is Such a People' (ITN) party during the pre-election period. By artificially inflating engagement metrics like shares and reactions, the network manipulated Facebook's algorithms to increase the visibility of posts by ITN politicians. The investigation confirmed the use of AI-generated imagery through tools like AIorNOT and PimEyes, highlighting a form of coordinated inauthentic behavior that violates platform policies and impacts democratic discourse.
A network of allegedly fake Facebook profiles using purportedly AI-generated profile images reportedly coordinated the spread of posts by members of Bulgaria's political party "There Is Such a People" (ITN) during the pre-election period. Reporting said the accounts repeatedly shared posts by ITN politician Pavela Mitova and ITN's leader Slavi Trifonov, helping make them appear highly engaging on Facebook and increasing their visibility.
Domain classification, causal taxonomy, severity scores, and national security assessments were LLM-classified and may contain errors.
Using AI systems to conduct large-scale disinformation campaigns, malicious surveillance, or targeted and sophisticated automated censorship and propaganda, with the aim of manipulating political processes, public opinion, and behavior.
Also classified as
The fake profiles were used to amplify specific false or misleading political claims, such as corruption allegations against a minister.
The incident involves a coordinated network of fake profiles using AI-generated images to conduct a political influence campaign aimed at manipulating public opinion and the democratic process.
Human
Due to a decision or action made by humans
Intentional
Due to an expected outcome from pursuing a goal
Post-deployment
Occurring after the AI model has been trained and deployed
The campaign was intentionally organized and executed by human actors using deployed AI image generation tools to create fake profiles.
Added categories peak at 3: Substantial (Epistemic Harm, Inferred)
A network of approximately 120 fake Facebook profiles in Bulgaria used AI-generated profile pictures to conduct a coordinated inauthentic behavior campaign, artificially amplifying content for a political party. While representing an established information operations tactic with minor national security impact, it highlights the ongoing use of synthetic media to target democratic discourse.
The AI systems involved (image generators) acted as tools to produce static synthetic images on demand, without exhibiting autonomous planning, decision-making, or interactive capabilities.
Limited Risk: The report describes the use of AI-generated content (deepfakes/synthetic profile images), which falls under transparency obligations to ensure users are aware they are interacting with or viewing AI-generated content.
The primary use of AI in the incident was generating deepfake profile images to make fake accounts look authentic. AI verification tools (AIorNOT, PimEyes) were used to detect them.
The report from Factcheck.bg is highly detailed, providing specific numbers of fake profiles (120 and 50), naming the targeted politicians, and detailing the verification methods (AIorNOT, PimEyes) used to confirm the AI-generated nature of the profile pictures. The facts are clear and consistent.
None. The coordinated inauthentic behavior and use of synthetic media are clearly documented.
Technology
AI-generated Profile Pictures
Fake profiles used AI-generated photos to look real.
AI-generated Text Descriptions
Russian descriptions suggest AI was used to write bios.
Algorithmic Recommendation Bias
Facebook algorithm prioritized posts with high fake engagement.
Data Inputs
Lack of Image Verification Data
No training data or checks to flag AI-synthesized faces.
Stolen Identity Data
Photos of real people were stolen to create deceptive accounts.
Human Factors
Hybrid Manual-Automated Control
Operators manually accepted friend requests to mimic real users.
Voter Deception
Bulgarian citizens were misled by fake AI-backed political posts.
Process and Methods
Coordinated Sharing Campaigns
Networks of fake profiles coordinated to share specific ITN posts.
Purchased Social Reactions
Buying reactions from foreign accounts to artificially boost reach.
Management
Inadequate Meta Moderation
Meta failed to proactively block coordinated inauthentic networks.
No Proactive AI Detection
Platform failed to integrate AI detection tools like AIorNOT.
Regulatory Environment
Political Advertising Bans
Bans on political ads forced reliance on organic reach manipulation.
Inadequate DSA Enforcement
EU Digital Services Act mandates are not fully stopping fake networks.
AI system involved
Identified by automated extraction; can differ from the alleged developers above.
