This is a research prototype. The data and analyses are preliminary and not yet validated — we'd welcome your .
A network of 171 bot accounts on X used ChatGPT to generate and spread pro-government political propaganda and defame the opposition candidate ahead of the presidential election in Ghana.
A network of 171 bot accounts on X are alleged to have used ChatGPT to generate political content supporting Ghana’s New Patriotic Party (NPP) and its presidential candidate, Mahamudu Bawumia, ahead of the December 2024 election. The AI-generated posts reportedly praised Bawumia while spreading disinformation targeting the opposition candidate, John Mahama, of the National Democratic Congress (NDC).
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 bot network generated false and disparaging claims accusing the opposition candidate of being a drunkard.
The campaign was enabled by X's failure to maintain adequate content moderation and trust and safety oversight.
The incident involved a coordinated network of AI-powered bots used to conduct a political influence campaign to manipulate public opinion and democratic processes in Ghana.
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 risk was caused by human bad actors who intentionally deployed a network of ChatGPT-powered bots to influence the Ghanaian election after the AI model was deployed.
Added categories peak at 4: Severe (Epistemic Harm, Inferred)
A network of 171 bot accounts on X used ChatGPT to generate and spread pro-government propaganda and defame an opposition candidate ahead of Ghana's presidential election. While highlighting the growing ease of deploying AI for localized political influence, the incident represents a minor national security threat with no compromise of physical systems, critical infrastructure, or core government operations.
The report describes ChatGPT being used to write social media posts favorable to a political party, acting as a writing assistant to the human operators of the bot network.
Risk Level 3: Limited Risk. The system involved is ChatGPT, a chatbot used to generate text content, which falls under the category of AI systems that require transparency obligations so users are aware they are interacting with AI-generated content.
The AI system (ChatGPT) was used to generate social media posts and political propaganda to influence the Ghanaian election.
The report is highly detailed, citing a specific investigation by NewsGuard that identified 171 bot accounts on X using ChatGPT. The role of the AI system (ChatGPT) is explicitly stated, and the nature of the campaign (pro-NPP, anti-NDC) is well-documented with examples of posts and hashtags. There is little ambiguity about the events, though the exact real-world impact on voters is noted as difficult to measure.
None
Technology
ChatGPT Text Generation
Allowed easy, automated creation of realistic political posts at scale.
AI-Generated Profile Photos
Created fake human personas to make bot accounts look authentic.
Algorithmic Amplification on X
X's algorithms prioritized verified bot content for higher visibility.
Data Inputs
Existing Political Narratives
Input of existing party talking points allowed bots to mimic real debates.
Lack of Content Labeling on X
Absence of labels identifying AI-generated content or automated accounts.
Human Factors
Public Trust in Blue Checks
Users believed blue checkmarks indicated reliable and verified accounts.
Bad Actors' Intent to Influence
Malicious actors deliberately leveraged AI tools to manipulate elections.
User Susceptibility
Users struggled to distinguish AI-generated bot accounts from real people.
Process and Methods
Diminished Content Moderation
X reduced trust, safety, and election integrity teams, weakening defense.
Predictable Posting Patterns
Bots posted at highly regular intervals, revealing automated behavior.
Ineffective Reporting Mechanisms
Slow response from platforms to reports of coordinated bot activity.
Management
Prioritization of Cost Cuts
X management cut 80% of safety engineers, reducing moderation capacity.
Monetization of Verification
Selling blue checks allowed bad actors to boost bot visibility.
Inadequate Abuse Prevention
OpenAI and X failed to proactively prevent political influence campaigns.
Regulatory Environment
Lack of AI Content Regulations
No legal frameworks governing AI-generated content in political campaigns.
Weak Platform Oversight
Regulatory bodies lacked authority to enforce election integrity on X.
AI system involved
Identified by automated extraction; can differ from the alleged developers above.
