By 
Artificial intelligence company Anthropic revealed that an individual in Kenya used its chatbot, Claude, to flood social media with fake political messages ahead of the country’s 2027 general election.
The operation generated content in sets of 50 posts at a time.
The operator explicitly instructed the AI to write in a casual tone so the posts would look like genuine, spontaneous opinions from everyday citizens instead of an organized propaganda campaign.
A large portion of the messages praised Energy Cabinet Secretary Opiyo Wandayi for blocking a planned increase in electricity prices, using promotional tags such as #PowerReliefKE and #PoweringTheNewKenya.
Simultaneously, other posts claimed that the main opposition coalition was falling apart, specifically targeting key political figures including former president Uhuru Kenyatta and former deputy president Rigathi Gachagua.
Anthropic stated that while the content strongly supported the ruling political side, there is no evidence that the Kenyan government ordered, funded, or participated in the project. Instead, the company described it as a domestic operation where a local user provided the political ideas, viewpoints, and hashtags, relying on AI simply to speed up the writing process.
The investigation also showed that the creator used this exact same setup under a separate commercial profile to write promotional posts for local retail businesses.
Despite the high volume of posts, the campaign failed to influence actual voters. Anthropic placed the effort at its lowest impact level, finding that the messages remained confined within a closed network of fake accounts and never reached or persuaded real people.
Kenya joins a growing list of countries where digital actors are testing AI tools for political influence. Similar covert campaigns using generative AI were recently detected and blocked in both Ghana and the Democratic Republic of Congo.
Comments