Anthropic: Claude AI Used for Kenya Political Posts Manipulation

• Anthropic's investigation revealed a pro-government operator in Kenya used Claude AI to generate fake social media posts ahead of the 2027 General Election.
• The AI-generated content praised Energy Cabinet Secretary Opiyo Wandayi and attacked opposition figures like former Deputy President Rigathi Gachagua and former President Uhuru Kenyatta.
• The operator instructed Claude AI to produce batches of 50 tweets designed to appear as spontaneous grassroots commentary.
• Anthropic classified the activity as Category One on its Breakout Scale, indicating it was isolated to fake accounts and failed to influence real people.
• While the campaign's tone suggested alignment with the ruling coalition, no evidence of direct government involvement was found.

This incident, involving Anthropic Claude AI Kenya political posts, underscores the emerging risk of AI-generated content in political campaigns and public discourse in Kenya.

A recent investigation by US-based artificial intelligence firm Anthropic has revealed a sophisticated attempt to manipulate public opinion in Kenya using its Claude AI platform. The company's report detailed how a pro-government operator, based within Kenya, leveraged the AI tool to generate and disseminate fabricated grassroots social media content. This activity was specifically designed to create the impression of spontaneous public support for certain political figures and narratives, particularly in the lead-up to the nation's 2027 General Election.

The operator utilized Claude AI across multiple sessions, providing explicit instructions to produce batches of precisely 50 tweets. A key directive was to ensure these posts appeared as "spontaneous grassroots commentary rather than a coordinated campaign." Many of the resulting AI-generated tweets lauded Energy Cabinet Secretary Opiyo Wandayi, specifically crediting him with preventing a scheduled electricity tariff increase by Kenya Power. To amplify their reach, these posts incorporated specific hashtags, including #PowerReliefKE and #PoweringTheNewKenya, on the social media platform X.

Beyond praising government figures, the AI-generated campaign also actively propagated narratives aimed at destabilizing the opposition. The operator pushed stories suggesting an imminent collapse of Kenya's United Opposition coalition ahead of the 2027 election. This effort directly targeted prominent opposition politicians, including former Deputy President Rigathi Gachagua, now a party leader, and former President Uhuru Kenyatta.

The targeting of Gachagua and other figures associated with the United Alternative Government coincided with their escalating criticism of President William Ruto's administration. President Ruto and his allies have consistently asserted that the opposition lacks a coherent plan or agenda for the country, focusing instead on critiquing the current government. Concurrently, retired President Kenyatta has faced significant backlash from President Ruto and his allies, drawn from the ruling UDA party and the opposition ODM party, which has a cooperation agreement with the government, and their affiliates, over his alleged continued involvement in political affairs. This criticism even led Nandi Senator Samson Cherargei to threaten sponsoring legislation that would strip Kenyatta of his retirement benefits due to his purported political activities. Furthermore, there have been claims that Kenyatta is financially supporting the Linda Mwananchi faction of the opposition, a group led by Nairobi Senator Edwin Sifuna and Siaya Governor James Orengo.

Anthropic's investigation, detailed in its Anthropic AI report Kenya, characterized the operation as a highly structured, manufactured grassroots campaign. It noted the consistent deployment of identical and unified messages regarding the electricity tariff increase across various online conversations. However, the company's assessment of the campaign's reach and influence indicated a limited impact. The activity was primarily confined to a network of fake accounts and local influencers operating on a single platform, ultimately failing to engage or sway any real individuals.

Employing its proprietary Breakout Scale, Anthropic classified the activity as Category One, signifying a low level of real-world impact. Crucially, the investigation found no evidence of government involvement in the operation, concluding that the activity appeared consistent with an entirely domestic Kenyan operation. While the tone of the campaign and the specific hashtags used suggested alignment with the ruling coalition, no direct government participation was substantiated. The investigation also uncovered that the same actor had previously used an identical AI workflow to promote Kenyan retail brands under a marketing persona identified as "SHANKI" or "Elkins Marketer," and that Kenyan brands were paying influencers to shape product narratives.

This incident, involving Anthropic Claude AI Kenya political posts, underscores the emerging risk of AI-generated content in political campaigns and public discourse in Kenya. The sophisticated use of Claude AI for political manipulation in Kenya, creating AI generated social media Kenya content, highlights a growing challenge for democratic processes. While this specific campaign's impact was limited, the potential for AI influence in the Kenya 2027 election and beyond remains a significant concern.

Lawyers and legal professionals should recognize this as a critical area for client advisement. Entities involved in elections, public relations, or any form of public communication must be vigilant in monitoring for AI-driven disinformation. Preparing for potential regulatory responses concerning AI use in political messaging is becoming increasingly vital as technology advances and bad actors seek new methods to shape public perception.

This incident underscores the emerging risk of AI-generated content in political campaigns and public discourse in Kenya. Lawyers should advise clients, particularly those involved in elections or public relations, on monitoring for AI-driven disinformation and preparing for potential regulatory responses concerning AI use in political messaging.

Source: Original reporting via The Standard