Shocking Revelation: AI Plot Unveiled to Disrupt Kenya's 2027 Elections

A covert artificial intelligence-driven influence operation targeted prominent Kenyan political figures, including former President Uhuru Kenyatta and former Deputy President Rigathi Gachagua, while simultaneously promoting Energy Cabinet Secretary Opiyo Wandayi. This sophisticated campaign, which aimed to sway public opinion ahead of Kenya's 2027 General Election, was uncovered and detailed in a threat intelligence report by the global AI firm Anthropic.

Identified as GTG-54004, the operation allegedly leveraged Anthropic's own Claude AI model to mass-produce political content. The primary objective was to cultivate the false impression of spontaneous grassroots support for or opposition to various political figures. According to the report, a single operator was responsible for using AI to generate a substantial volume of social media posts. These posts included narratives specifically designed to portray the United Opposition as fractured and on the verge of collapse as the 2027 elections approached.

The actor behind the campaign reportedly used Claude in multiple sessions to generate batches of exactly 50 tweets, explicitly instructing the AI model to craft posts that resembled spontaneous grassroots commentary rather than a centrally coordinated campaign. This approach was crucial for the campaign's efforts to target Gachagua and Kenyatta, aiming to highlight perceived divisions within the opposition movement and undermine their standing.

In addition to targeting opposition figures, the operation actively pushed favorable messaging regarding Energy Cabinet Secretary Opiyo Wandayi. A key focus of this positive narrative was Wandayi's role in halting a previously scheduled electricity tariff increase. The campaign utilized specific hashtags, such as #PowerReliefKE and #PoweringTheNewKenya, to disseminate these AI-generated posts, presenting them as genuine independent commentary from ordinary social media users.

The Anthropic report further revealed that the operator would supply Claude with pre-drafted topics, talking points, and hashtags, subsequently instructing the model to enhance the content's natural and human-like appearance. Interestingly, the same AI-assisted workflow employed for political influence was also allegedly used for commercial marketing campaigns. Operating under a marketing persona, 'SHANKI' or 'Elkins Marketer,' the actor ran identical AI workflows for Kenyan retail brands. This raises significant questions about the potential repurposing of techniques developed for advertising in the realm of political manipulation.

The influence operation utilized templates and consistent instructions across both its political and commercial endeavors. Posts were generated in batches and meticulously packaged, facilitating the simultaneous copying and publication of multiple messages. Anthropic stated that the primary intent of this activity was to manufacture the appearance of organic online engagement, rather than merely generating generic political content.

Despite the extensive scale of the AI-generated material, Anthropic assessed the operation's overall impact as minimal, classifying it as Category One on their Breakout Scale. This indicates that the activity was completely isolated within its network of fake accounts and local influencers on a single platform, ultimately failing to reach or influence any genuine individuals. The campaign did not manage to penetrate wider public discourse or meaningfully affect real users.

Anthropic's investigation found no evidence directly linking the operation to the Kenyan government or suggesting it was government-funded. Instead, the activity was assessed to be a domestic Kenyan influence operation. The network was initially identified following threat intelligence shared by OpenAI concerning suspected repeated misuse of AI systems. Following the discovery, Anthropic took immediate action, banning the involved accounts and deploying additional behavioral detection mechanisms to identify similar future activities.

These findings underscore the escalating use of generative AI in political communication, particularly its capacity to produce vast quantities of personalized messages at a speed and scale that would be unattainable for conventional human-driven campaigns. While this specific Kenyan operation did not achieve significant public reach, the report serves as a stark illustration of how AI tools can be deployed to blur the critical distinction between genuine grassroots political discussion and centrally coordinated, manufactured messaging, especially as political parties and movements gear up for the 2027 election.