A Kenyan lawsuit has put a little-known political analytics business at the centre of an urgent question: can a private firm quietly classify voters before an election? The claim, filed on Friday at the High Court in Nairobi, says Tawi Signal Analytics assigned “coordination risk” labels to 1.2 million profiles and marked 14 counties as high risk ahead of the 2027 vote.
The 23 plaintiffs — election monitors, shop owners and former contractors — want the court to halt the product’s use and order the firm to preserve its records. Their lead representative, Isaac Cardona-Solís, says the labels could brand ordinary political discussion as manipulation, affecting how campaigns spend money and which neighbourhoods they trust.
The score behind the suit
The suit cites a sales presentation describing a dashboard that draws on public posts and groups profiles into county-level risk ratings. It does not show that the ratings changed any campaign decisions, or establish that private messages were collected. Tawi Signal has not filed a defence; the court registry confirmed the claim was received.
The firm’s spokesperson, Margarette Okuneva, said in an interview that Tawi Signal sells “area-level trend analysis, not lists of people for campaigns to target”. She said the company had not built a database of political beliefs and would contest any claim that it had improperly handled personal information.
“A county label can become a permission slip to treat millions of people as suspect,” said Hilppa Arhinmäki, a Nairobi-based data-governance researcher who reviewed the sections of the filing made public. She warned that broad risk scores can be reused beyond their original purpose, while leaving the people described with no practical way to challenge them.
A county label can become a permission slip to treat millions of people as suspect.
Hilppa Arhinmäki, data-governance researcher
The plaintiffs seek damages of KSh2.4 billion (about €16 million) and an interim order barring Tawi Signal from selling or updating the ratings. Their lawyer argues that Kenya’s Data Protection Act gives people rights over information used to make decisions about them, but the suit does not yet include a technical audit of the scoring system.
A vote approaching fast
That gap is central to the alarm. Abel van Wijk, an election-technology specialist who has advised monitoring groups in East Africa, said a score can influence decisions even when nobody treats it as a formal instruction. “If campaign teams decide a place is too risky to engage, the residents pay the price without ever seeing the label,” he said.
Kenya’s next general election is scheduled for 2027, and parties are already building voter-outreach plans. A ruling on access to the scoring method could clarify whether the firm must explain how its ratings are produced; a temporary ban, by contrast, could arrive too late to answer whether labels have already circulated among clients.
The court has not set a hearing date. For now, the plaintiffs’ filing offers allegations, not proof that any voter was excluded or that the ratings were used to influence a campaign. But the dispute has moved a question usually buried in software contracts into open court: who gets to decide when political speech looks suspicious?
