Urgent.News

What's breaking now, across thousands of outlets.

AI

Kenya’s AI Gamble: Can Nairobi harness Anthropic without losing control of its data?

Kenya is moving to deepen its use of artificial intelligence while drawing a line around one of the most sensitive issues in the digital economy: who controls the data and systems on which AI increasingly depends. A new cooperation framework between the government and US-based artificial intelligence company Anthropic could open the door to AI […]

Kenya is seeking to leverage artificial intelligence (AI) while safeguarding control over its data, as highlighted by a new cooperation framework with US-based AI company Anthropic. Signed on September 22, 2026, the Joint Declaration sets the stage for AI applications in health, education, and public services, but success hinges on Kenya's ability to develop locally managed systems.

The agreement focuses on AI skills, research, responsible public-sector applications, safety, and evaluation, as well as locally relevant applications.

Unlike a one-time technology deployment, the declaration serves as a platform for individual ministries, departments, and agencies to negotiate more specific programs with Anthropic. This flexible structure will be critical to implementation. Data governance is identified as a boundary that cooperation should not cross, as any future program involving data access, processing, or use will be governed by Kenyan law, covering data protection, cybersecurity, and public-sector requirements.

This provision is vital because AI expansion in government might involve sensitive information held by public institutions.

Kenya is not only concerned with maintaining control over data but also aims to play a role in the development, evaluation, and governance of advanced AI systems. The declaration emphasizes strengthening African capacity in these areas, including local research and technical and institutional capacity. Moreover, AI systems deployed in Kenya should reflect the country's languages, institutions, development priorities, and social context.

This requirement could become especially important as AI tools transition from general-purpose applications to public services.

Kenya faces a dual challenge: acquiring advanced AI capabilities while simultaneously building sufficient domestic expertise to scrutinize, adapt, and govern those technologies. The agreement has yet to outline these outcomes, but its subsequent phase will involve pilots, institutional partnerships, research initiatives, and capacity-building programs.

Dr. Abraham Korir Sing’Oei, Principal Secretary for Foreign Affairs, and Ambassador Philip Thigo, Special Envoy on Technology, signed the declaration, viewing it as part of a broader technology diplomacy strategy. However, the true evaluation will not be based on the number of AI partnerships Kenya establishes. Instead, it will focus on whether these collaborations result in enhanced researchers, improved public services, increased technical capacity, and AI systems that Kenya can comprehend, evaluate, and govern.

Kenya's AI strategy is now transitioning to a more challenging phase: transitioning from attracting global technology to developing the domestic capacity necessary to manage how this technology is utilized.

Written by urgent.news from People Daily Kenya's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at peopledaily.digital →

More in AI

A 1.5MB Local LLM Manager That Turns llama.cpp Into a Complete Desktop AI Workstation

Open Source | Ooor: A 1.5MB Local LLM Manager That Turns llama.cpp Into a Complete Desktop AI Workstation TL;DR : Ooor is a MIT-licensed, open-source Windows desktop app weighing just 1.5MB.

  • Ooor is a 1.5MB Windows desktop app managing llama.cpp engine
  • Pure local solution with no Electron, cloud services, telemetry, or accounts
  • Integrates Llama Engine Manager, Model Library Manager, Downloader, Chat Console & Agent

Agentic AI vs AI Agents: Orchestration Beats a Lone Worker

An AI agent wins when the job is a single, well-scoped, verifiable task: one worker, one toolset, one output you can check.

  • Agentic AI differs from AI agents in scope and coordination capabilities.
  • Building a single agent first provides baseline for scoring and troubleshooting.
  • Orchestration in agentic AI introduces complexity with routing, shared state, and partial failures.

Claude Opus 5.5: 40% cheaper, frontier-grade performance

This week was dominated by cost compression and gateway consolidation. Claude Opus 5.5 dropped with a meaningful price cut and no code changes required, while Vercel's AI Gateway absorbed four new…

  • Claude Opus 5.5 released with 40% cost reduction
  • Frontier performance comparable to Fable 5.1
  • Cache reads reduced by 60% to $0.20 per million tokens

More from Thursday 24 September →