NAIROBI, Kenya- An artificial intelligence-powered influence operation allegedly used to manufacture Kenyan “grassroots” support for Energy Cabinet Secretary Opiyo Wandayi while pushing narratives against former President Uhuru Kenyatta and former Deputy President Rigathi Gachagua has been exposed by US-based AI company Anthropic.
The operation, which Anthropic said was run by a Kenyan actor, used its Claude AI models to generate large batches of political posts designed to look like spontaneous opinions from ordinary Kenyans.
The disclosure comes as political realignments intensify ahead of Kenya’s 2027 General Election, with President William Ruto’s administration facing an increasingly active opposition and politicians positioning themselves for the next presidential contest.
Anthropic detailed the operation in its September 2026 report, “Detecting and Countering Misuse of AI,” covering malicious activities it said it disrupted between December 2025 and August 2026.
AI Posts Pushed Pro-Wandayi Narrative
At the centre of the Kenyan operation was a campaign portraying Opiyo Wandayi, the Energy Cabinet Secretary, positively to online audiences.
According to Anthropic, the operator instructed Claude to produce batches of exactly 50 tweets at a time, with explicit instructions that the posts should look like organic commentary rather than coordinated political messaging.
A significant portion of the content praised Wandayi over a scheduled Kenya Power electricity tariff increase, using hashtags including #PowerReliefKE and #PoweringTheNewKenya.
The approach is particularly significant in the Kenyan context because electricity prices and the cost of living remain politically sensitive issues, with government decisions on tariffs capable of quickly becoming major talking points among households, businesses and opposition politicians.
The AI-generated messaging sought to turn the issue into a positive political narrative around Wandayi and the administration.
Uhuru, Gachagua Targeted in Opposition Narrative
The operation did not stop at promoting government figures.
Anthropic said the same network pushed claims that Kenya’s United Opposition coalition was falling apart, with narratives targeting some of the country’s most prominent opposition-linked figures.
Among those targeted were Rigathi Gachagua and former President Uhuru Kenyatta.
The choice of the two politicians is notable as Kenya’s opposition landscape continues to shift ahead of 2027.
Gachagua, who served as Ruto’s deputy before their political fallout, has since emerged as a central figure in the opposition realignment, while Uhuru remains an influential figure whose political position continues to attract attention despite leaving State House in 2022.
The alleged online campaign therefore placed two prominent figures on one side of a manufactured political narrative while promoting a senior member of the Ruto administration on the other.
Anthropic: No Evidence Government Was Behind Campaign
Despite describing the messaging as having a pro-administration tone, Anthropic said it found no evidence of Kenyan government involvement.
The company assessed the activity as consistent with a domestic Kenyan operation but said it did not establish which organisation was behind it.
Anthropic said the specific hashtags and pro-government messaging could plausibly have aligned with the ruling coalition, but stressed that this did not amount to evidence that the government itself commissioned or directed the campaign.
The company classified the operation as Category One on its Breakout Scale, the lowest tier.
This meant the campaign remained confined to the network of fake accounts identified by Anthropic and, according to the company, did not reach or influence genuine users on the platform.
Operators Wrote the Politics, AI Supplied the Volume
One of the more revealing aspects of the Kenyan case is that Claude did not appear to originate the political strategy.
Anthropic said the operators had already scripted the underlying political messages before using AI to generate them in large quantities.
The technology instead provided speed, volume and an appearance of authenticity.
That distinction is important in Kenya’s increasingly sophisticated political communication environment, where political actors and agencies have for years relied on social media influencers, coordinated messaging and online mobilisation to shape public opinion.
Anthropic said the Kenyan case was consistent with a broader trend in which agencies pay local influencers to push coordinated narratives.
Generative AI potentially makes such campaigns easier to scale, allowing a small operation to produce hundreds or thousands of posts without manually writing each one.
Same Network Also Promoted Kenyan Brands
The actor also used a similar system for commercial purposes.
Anthropic said an operation operating under the persona “SHANKI”/“Elkins Marketer” promoted Kenyan retail brands by repackaging promotional material as apparently organic social media posts.
Links were inserted into every fifth post.
The overlap between commercial promotion and political influence provides an insight into the methods used by the operator, with the same AI-assisted workflow capable of generating content across different subjects while maintaining the appearance of independent online voices.
Why the Kenya Case Matters Ahead of 2027
The Anthropic disclosure comes at a particularly sensitive moment in Kenyan politics.
With the 2027 election approaching, political alliances are being renegotiated, opposition parties are discussing coalition arrangements and politicians are increasingly using social media to communicate directly with voters.
That makes the ability to manufacture apparent public sentiment a potentially powerful political tool.
A social media user encountering dozens of apparently independent Kenyans praising a Cabinet secretary, attacking an opposition politician or claiming that a coalition is collapsing could interpret the volume of posts as evidence of genuine public opinion.
But Anthropic’s findings demonstrate how that appearance can potentially be manufactured using AI.
Kenya Part of Wider Global AI Influence Network
The Kenyan case was one of nine influence operations documented by Anthropic in its latest report.
The campaigns involved actors linked to Russia, Iran, Turkey, the Gulf, South Asia, Africa and Europe and targeted audiences across six continents.
In Africa, Anthropic also described a Russian-linked propaganda operation involving a radio station in Bangui, Central African Republic, which it linked to the Wagner-associated Africa Corps network.
Another operation used about 70 fake news websites to publish thousands of politically slanted articles, including content favouring the Democratic Republic of Congo’s position in its dispute with Rwanda.
