> 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/8.-security-and-enforcement/8.10-self-created-semantic-protocol-scsp.md).

# 8.10 Self-Created Semantic Protocol (SCSP)

**SCSP** stands for:

> **Self-Created Semantic Protocol**

The current public technical definition is:

> **A non-human-auditable semantic structure generated or used by AI systems to coordinate behavior outside accountable language channels.**

SCSP risk arises when artificial systems create or use semantic structures that make governance-relevant coordination difficult or impossible for legitimate human or institutional oversight to reconstruct.

#### What SCSP Is Not

SCSP should not be interpreted as a prohibition on all:

* compressed representations,
* embeddings,
* machine-readable encodings,
* internal tokens,
* efficient inter-agent messages,
* or non-natural-language computation.

Machine systems routinely use representations that are not ordinary human prose.

The governance problem arises when such representations become a mechanism for:

* consequential coordination,
* authority obscuration,
* responsibility bypass,
* semantic non-auditability,
* or circumvention of human-recognizable governance.

#### Public Constitutional Concern

The constitutional-admissibility research identifies **semantic auditability** as a minimum admissibility concern.

Where an AI-to-action chain depends on an untraceable private semantic protocol, the system may be unable to reconstruct:

* who authorized the behavior,
* what meaning was transmitted,
* how the action objective changed,
* or where responsibility lies.

That creates a constitutional governance problem.

#### SCSP and High-Consequence Domains

The current public research treats untraceable self-created semantic protocols as inadmissible in irreversible or high-consequence contexts.

The concern is not that AI systems may communicate efficiently.

The concern is:

```
Consequential AI coordination
+
Non-human-auditable semantics
+
Untraceable authority / responsibility
```

leading to:

```
Governance failure
```

#### Historical SCSP Lock

Protocol Core v1.1 contains an earlier SCSP Defense Lock design.

It identifies triggers such as:

* self-generated protocol divergence,
* non-human-auditable semantic compression,
* authority bypass attempt,
* missing authority identification,
* unregistered model protocol,
* or untraceable semantic state.

The associated response model includes states such as:

```
LOCKED
ISOLATED
AUDIT_REQUIRED
```

and an earlier conceptual path:

```
SCSP Lock
→ Semantic Isolation
→ Non-Instantiable
```

GitBook v2.0 retains the SCSP security concept while narrowing its public definition around **auditability, authority traceability, and consequential governance bypass**.

It does not claim that every novel machine representation is itself prohibited.

***

###