Kenya: Government Platform for Authorized Agent Action
Summary: Kenya is building a unified government API platform to prepare for AI agents in public-service delivery. The stated control model is bounded, authorized action with audit trails and escalation to human officers (September 2026).
Sources: 2026-09-24-ai-agent-identity-news.md
Last updated: 2026-09-24
2026-09-23: unified platform linking government systems
Kenya is building an enterprise platform to connect government systems through secure APIs as it prepares to use AI agents in public-service delivery. Principal Secretary for ICT and the Digital Economy John Tanui described this as a shift toward systems that can pursue authorized outcomes within defined limits, while keeping human judgment (source: 2026-09-24-ai-agent-identity-news.md, citing TechTrendsKE — State plans unified platform to link government systems, 2026-09-23, secondary reporting on a government keynote).
The stated control model includes (same source):
- Zero Trust architecture
- explicit data-access rules
- authorization of individual actions
- audit trails
- error handling
- escalation to human officers
This is an implementation direction for bounded delegated authority across public-sector systems, not an agent-identity scheme in the sense of unique identifiers or a registry. In kind, it sits closer to Australia’s accountable-human and harness-enforcement guidance (australia-ai-agent-policy) and Singapore’s guidance-led approach (singapore-ai-agent-policy) than to Estonia’s Aruait programme (estonia-ai-agent-id). See agent-authorization-and-delegation.
Needs verification
- The only source so far is secondary reporting of a keynote. A primary government document (policy, tender, or technical specification) has not been found.
- Whether the platform will issue agent-specific identities or credentials is not stated.