Artificial intelligence (AI) company Anthropic has revealed how its Claude chatbot was used to mass-produce fake social media posts praising Energy Cabinet Secretary Opiyo Wandayi and attack Kenya’s opposition.
The AI safety and research firm said a Claude user generated Kenyan political content designed to appear as organic public sentiment, including batches of fake posts on the social media platform X.
The content praised Mr Wandayi for allegedly stopping a scheduled electricity tariff increase.
The operation also pushed claims that the United Opposition coalition was breaking apart ahead of the 2027 elections, targeting politicians including former deputy president Rigathi Gachagua and former president Uhuru Kenyatta.
It reveals how AI is emerging as a new tool for political influence in Kenya ahead of the 2027 General Election, as political operators use the technology to manufacture apparent public support for politicians.
‘We identified and removed an account that was used by a single actor to mass-produce Kenyan political content,’ Anthropic said in a new report detailing AI misuse cases on its platform between December 2025 and August 2026.
‘The operation was built to look like spontaneous, grassroots public sentiment. The actor used Claude across several sessions to generate batches of exactly 50 tweets.’
The Claude prompts instructed the chatbot that the posts should look like organic grassroots commentary rather than a coordinated campaign. They used hashtags including #PowerReliefKE and #PoweringTheNewKenya to amplify the messaging.
While Anthropic’s investigation did not establish the identities of those behind the operation, the company said the pro-government tone and specific hashtags suggested that it was plausibly aligned with President William Ruto’s supporters.
‘We believe this was a local Kenyan political astroturfing campaign,’ the report said, referencing an organised effort that looks like a natural, spontaneous public movement, but is sponsored by hidden political interests.
‘The network’s pro-administration tone and specific hashtags suggest it was plausibly aligned with the ruling coalition, though we have not identified the exact organisation responsible.’
Read: Protecting your reputation in age of AI, and fake news
It said the operation did not ultimately reach real people, with its activity remaining isolated within a network of fake accounts and local influences on a single platform.
Anthropic said it probed Kenya following a tip from the rival US firm OpenAI about undisclosed criminal activity on its ChatGPT platform.
‘While a social media site usually sees an operation once its content is already circulating, we may see it on Claude while the operation is still being built,’ the company said.
The investigation comes amid growing concern about the use of AI to manipulate political and civic discourse.
Kenya’s information ecosystem is increasingly dependent on social media platforms such as Facebook, X, Instagram, TikTok, YouTube and WhatsApp, creating new avenues for political messaging outside traditional platforms like television, newspapers and radio.
A recent Kenya ICT Action Network (KICTANet) study found that alternative media actors on social media are shaping political narratives, often with less editorial accountability than traditional media.
The study, based on monitoring conducted between January and March 2026, found a pattern in which misleading narratives can move across platforms and become progressively harder to challenge.
The think tank flagged Facebook for providing the starting point through fabricated newspaper headlines, doctored political graphics, and impersonated editorial content.
It said TikTok then accelerates the narratives through emotionally charged and selectively edited clips, while YouTube and podcast channels give them a longer life through extended commentary and speculation on platforms like podcasts.
‘Platform algorithms play an important role… Content that generates stronger emotional reactions-anger, humour, outrage, fear, or political loyalty-is often amplified more widely through recommendation systems and reposting behaviour,’ said the report.
‘As a result, sensational political narratives may receive greater visibility than balanced or evidence-based reporting.’
Read: How Raila deepfake exposes digital weak spot, sparks AI alarm
The growing use of generative AI raises the risks, as experts warn that AI-generated deepfakes and manipulated content will become more sophisticated ahead of the 2027 elections.
‘False content may become more convincing and harder to detect; manipulated media may spread rapidly across platforms; and verification processes may struggle to keep pace,’ KICTANet said.
Kenya has faced politically driven online manipulation before. During the 2017 General Election, British political consulting firm Cambridge Analytica was embroiled in controversy over claims that it used data analytics, psychological profiling, social media ads and grassroots networks to influence voters.
The company was accused of using Facebook user data obtained through a third-party application to target users with fake-news campaigns.