AI Governance: Featured Story
Coverage from Games Beat, Hpcwire, and others
Kihan Launches Constitutional Computing for AI

Kihan announced Constitutional Computing, an architecture that checks consequential AI agent actions against signed rules independently of the AI model. Its Inktomi decision engine can permit, refuse, or escalate requests and produce verifiable decision records, but production enforcement integrations remain under development. Kihan is targeting regulated industries and plans initial enterprise evaluations for December 2026.
If you read one thing
Provides the clearest overview of Inktomi’s authorization design, current integration status and planned enterprise evaluations.
The evidence
Adds a concise account of the launch while noting that implementation details, deployment results and independent evaluations were not provided.
The decision engine is reported operational, but production enforcement integrations are still in development. Kihan is targeting initial enterprise evaluations for December 2026; the supplied coverage does not establish deployment outcomes or independent validation.
Kihan says Inktomi checks proposed actions against signed rules and can permit, refuse or refer requests for human approval, with verifiable decision records.
The company reports that Inktomi’s decision engine is operational, while production enforcement integrations remain under development.
Kihan targets initial enterprise evaluations in December 2026 and is focusing on regulated sectors including finance, insurance, healthcare and legal services.
The coverage reports the architecture’s intended design and development status, but provides no deployment results or independent evaluation of its effectiveness.
