Who Gave the Order? When AI Issues Commands Without a Speaker

A reflection on how artificial language exerts control through structure, not speech

1. What This Article Examines

Consider the sentence:

"All credentials must be verified before access is granted."

It appears normal. It sounds official. But a basic question remains:

• Who is supposed to verify them?


• Who gave the instruction?

This question anchors the published article Syntax Without Subject. The study examines how artificial intelligence, particularly large language models (LLMs), generates institutional documents that impose conditions, deliver obligations, or regulate behavior, yet do so without identifying any speaker or responsible agent.

The language operates correctly. However, no one appears to be speaking.

**2. What the Corpus Shows

**A structured analysis was conducted on 172 AI-generated documents across legal, medical, and administrative contexts. These were not fictional samples. They included actual policy drafts, contract templates, and internal protocols.

Three syntactic patterns appear consistently:

Passive constructions, such as “Data shall be retained,” which avoid naming the actor.

Nominalizations, such as “submission,” which replace verbs like “submit” to obscure who acts.

Instruction templates without subjects, such as “Confirm identity before approval,” which issue commands without specifying any agent.

Each of these forms is grammatically acceptable and often institutionally adopted. What unites them is that they structurally remove the subject, the speaker, and the agent.

3. Why This Has Consequences

Where no one is named, no one is held accountable.

In legal and institutional systems, responsibility depends on attribution. Clauses without identifiable subjects weaken the enforceability of directives and create space for untraceable execution.

The article introduces the concept of the traceability threshold. This threshold marks the moment when a sentence retains formal authority but no longer contains any referential path to a speaker, issuer, or author. What remains is structurally executable, yet syntactically anonymous.

Beyond this point, grammar does not describe or deliberate. It acts.

**4. What This Reveals About AI-Governed Language

**As institutions begin to adopt AI-generated rules, a shift occurs. Directives are no longer authored. They are compiled. The structure replaces the source. Compliance is triggered by form, not by communicative intent.

This development does not resemble censorship. It resembles formatting.

Some of the most powerful language is not that which commands loudly, but that which silently removes the actor. Formality, once a guarantee of procedure, becomes a mask for the absence of responsibility.

Documents continue to function. Authority continues to be exercised. Yet no one signs, and no one speaks.

🔗 Read the Full Article

Syntax Without Subject: Structural Delegation and the Disappearance of Political Agency in LLM-Governed Contexts

Agustin V. Startari

Linguist and researcher in historical and artificial language systems

ORCID: https://orcid.org/0009-0001-4714-6539

*SSRN Author ID: *https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5370857

Ethos

I do not use artificial intelligence to write what I do not know. I use it to challenge what I do. I write to reclaim the voice in an age of automated neutrality. My work is not outsourced. It is authored.

— Agustin V. Startari