> For the complete documentation index, see [llms.txt](https://docs.lpp-minduniverse.org/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.lpp-minduniverse.org/lingua-pactum-protocol-lpp-documentation/1.constitutional-foundation/1.4-responsibility-and-attribution.md).

# 1.4 Responsibility & Attribution

LPP treats responsibility as an upstream governance property.

Responsibility is not satisfied merely by producing a log after an action has occurred.

The relevant chain must remain attributable from language and intent through execution and evidence.

#### Semantic Responsibility

AI-generated language may influence or directly form consequential action.

For this reason, semantic transformation cannot be treated as responsibility-neutral.

A responsibility chain may need to preserve the relationship between:

```
Human / Institutional Authority
        ↓
Authorized Intent
        ↓
AI-Generated Statement
        ↓
Task / Delegation
        ↓
Action Candidate
        ↓
Execution
        ↓
Outcome
```

The key constitutional question is whether responsibility remains identifiable across transformation.

A statement that materially changes meaning while preserving only superficial textual or procedural continuity may no longer preserve the same responsibility basis.

This becomes particularly important in multi-agent systems, where an initial instruction may be:

* decomposed,
* rewritten,
* summarized,
* delegated,
* optimized,
* transformed,
* or combined with additional context.

Responsibility must survive these transformations.

#### Action Responsibility

When language crosses into action, responsibility cannot terminate at:

* “the model decided,”
* “the agent selected the tool,”
* “the automation executed,”
* or “the system produced the result.”

These describe mechanisms.

They do not identify the legitimate bearer of authority and responsibility.

Article IV therefore requires accountability to terminate at an identifiable:

* natural person, or
* legal entity.

This prevents responsibility from disappearing into technical abstraction.

#### Traceability

Traceability is the ability to reconstruct the relevant governance lineage.

For consequential actions, this may include:

* authority origin,
* authorized intent,
* scope,
* responsible principal,
* semantic transformations,
* delegation events,
* admission decisions,
* Permit issuance,
* execution,
* and resulting evidence.

Traceability does not mean that every internal model computation must be fully exposed.

The constitutional requirement concerns the governance-relevant chain.

The system must preserve enough information to determine:

> **Who authorized what, what changed, what was executed, and under what legitimate basis?**

#### Responsibility Vacuum

A responsibility vacuum exists when an AI-triggered action has material consequence but no identifiable human or institutional endpoint of accountability.

Under the constitutional model, this is not a neutral documentation deficiency.

It is a legitimacy failure.

***