The players

An AI action rarely belongs to one organization. Responsibility can still scatter across all of them.

Sources and reading context

TENIR Labs · editorial responsibility map · not legal advice

Simplified public rendering, read on 19 October 2026. KB3: enacted, active doctrinal rendering line. It renders meaning; it is not the code, and not a constitution. KB2: prior rendering line. This site certifies no implementation. Read the detailed status →

A fragmented chain of responsibility

An AI action crosses a market: model, platform, data, integration, business rules and deployment decisions. Each actor knows its commercial perimeter; consequences cross the borders between them.

Player	What they promise	What they should clarify
Model vendor	Performance and autonomy	Version, limits and the boundary between model capability and system guarantee
Platform provider	Orchestration and integration	Services, permissions and responsibilities left to the customer
Integrator	Fast deployment	Configuration, connectors, tests and introduced assumptions
Customer organization	A reliable service or decision	Mandate, limits, supervision and reconstruction capability
Business team	An operational result	Rules, exceptions and decisions not implicitly delegated
DSI / security	Infrastructure and protection	Available controls, access, logs and effective stop conditions
Affected person	Receives the effect	The right to understand, challenge and obtain an answer

Five questions

What exactly is being promised?
What is actually controlled?
What is not controlled?
What evidence should exist?
Which responsibility may be sent elsewhere?

Distributed responsibility is not absent responsibility. It must be reconstructed before the incident, not invented after it.

Replay a case sequence →

Responsibility map

One consequence. Several responsibility zones.

Each player owns one piece of the chain. None should own only the commercial story.