AS Ghana approaches its crucial presidential election on December 7, a significant social media manipulation effort has come to light. Research by NewsGuard, a website that provides tools to evaluate the accuracy and trustworthiness of news sources, has uncovered a network of 171 AI-powered bot accounts on X (formerly Twitter) designed to influence the political discourse in favour of the ruling New Patriotic Party (NPP) and its presidential candidate Mahamudu Bawumia. The bots are also targeting opposition leaders, particularly John Mahama of the National Democratic Congress (NDC), through derogatory messages and disinformation.
The bot network, which has been active since February 2024, has been promoting Bawumia’s political agenda and policies using a range of AI-generated content. The posts typically use popular NPP hashtags such as #Bawumia2024 and #ItIsPossible, with the bots consistently framing Bawumia as a transformative leader for Ghana. One example from September reads, ‘Admiring Dr. Bawumia’s commitment to transforming Ghana,’ with further posts echoing similar positive messaging about the candidate.
In addition to promoting Bawumia, the bots have been actively attacking Mahama, with posts disparaging him and calling him names like ‘Drunkmani Mahama’ and ‘mahamaisaliar.’ These derogatory labels aim to damage Mahama’s reputation, despite his denial of such accusations. This targeted smear campaign is a clear strategy to discredit the NDC and its candidate, influencing public perception ahead of the election.
‘The primary goal of the network appears to be to amplify pro-NPP messaging, promote the Bawumia administration and take aim at the opposition National Democratic Congress,’ McKenzie Sadeghi, an editor of AI and foreign influence at NewsGuard who contributed to the research, told Rest of World, a nonprofit publication that challenges assumptions about whose experiences with technology are valued.
The AI-driven nature of these bots is one of the most concerning aspects of the research. The profiles of the accounts appear artificially created, with AI-generated images and generic names such as ‘Glenn Washington’ and ‘Netflix Series&Movies.’ These bots post at regular, predictable intervals—often as many as 10 times a day—indicating a highly coordinated effort. This type of regular posting and engagement, often limited to the bot network itself, points to the automated nature of the activity.
NewsGuard researchers used Pangram Labs’ AI detection tool to analyse the posts from these accounts, and the results confirmed that the majority of the content was generated by ChatGPT, an AI language model developed by OpenAI. The scale at which these bots produce content has raised alarms about the ease with which AI can now be used to manipulate public discourse and shape political outcomes.
The timing of this discovery is especially significant given the current state of content moderation on X. Since Elon Musk’s acquisition of the platform in 2022, there has been a significant reduction in staff dedicated to content moderation, with many experts warning that this has created opportunities for manipulation. Musk’s decision to cut the trust and safety team and eliminate the election integrity unit has left a vacuum that bad actors are increasingly exploiting.
Dimitris Dimitriadis, NewsGuard’s director of research and development, noted that the bot accounts are posting during peak hours in Ghana, between 8 a.m. and 6 p.m., suggesting that these posts are carefully timed to reach the maximum number of users. With minimal content moderation and little oversight of verified users, these bots are able to spread misinformation with greater ease than ever before. Research from NewsGuard also found that users with blue verification badges on X often amplify false or unsubstantiated claims, further enabling the spread of manipulated content.
This AI manipulation comes at a volatile time for Ghana, where the 2020 election cycle saw violence and fatalities, with eight people killed during the electoral process. There are concerns that the spread of AI-generated content could exacerbate tensions in what is already a high-stakes election. NewsGuard’s research suggests that while the bot network has primarily been used to promote NPP messages and smear opposition candidates, it’s difficult to assess the full extent of its impact on public opinion. With thousands of users posting similar content daily, distinguishing the influence of these bots from genuine political discourse remains a challenge.
The use of AI-driven bots in Ghana’s election marks a worrying development in the use of technology for political manipulation. Experts warn that without stronger regulations and greater transparency, AI could continue to play a growing role in influencing electoral outcomes worldwide. Dimitriadis from NewsGuard cautioned that AI is increasingly being used by bad actors to shape political discourse, a trend that could undermine the integrity of democratic processes.
Although some of the AI accounts have been suspended or restricted by X, at the time of writing, many of the bots remain active. NewsGuard shared their findings with X and OpenAI, but neither company has provided a detailed response to the allegations. OpenAI has previously acknowledged concerns over the use of its technology in political manipulation, with reports indicating that ChatGPT has been used to generate content for influence networks in several countries, including the US and Rwanda.
As Ghana’s election approaches, the continued use of AI-driven bot accounts to sway public opinion remains a critical issue. The case highlights the need for stronger safeguards to protect the integrity of elections and prevent AI from being used to manipulate democratic outcomes.
