How the Claudia-Lumus Protocols Forge a Radical Craft in the Posthuman Era

Much has been said about the future of professions in the age of artificial intelligence. Some fear their disappearance. Others, their trivialization. But few have paid attention to a less spectacular, yet deeper fact: work can not only survive AI, it can be redefined from within, with new principles, new forms, and new functions.

That is what has happened to my craft. I did not protect it from AI. I did not delegate it either. I redefined it as a space of assisted, but not automated, thought. A radical worksite where language remains a political frontier.

What I call the Claudia-Lumus Protocols are not mere style rules or user preferences. They are operational structures that govern the form, substance, and method of interaction with a general-use AI, demanding rigor, ethical consistency, traceability, and respect for the semantic and political hierarchies of language.

I. The difference between a prompt and a protocol

A prompt is a one-time instruction. It works as a direct command that produces a result: a phrase, a summary, a translation, a stylistic imitation. It is the classic mode through which millions of users interact with AI systems: immediate, localized, frameworkless.

A protocol, on the other hand, is a structure of discursive governance. It does not merely request something. It defines the world in which that something makes sense.

In my case, the protocols I have designed —and which the AI has been trained by me to obey— regulate not only tone or format. They regulate:

  • The method for verifying facts under information censorship.
  • The criteria for distinguishing between primary sources, intermediaries, and editorial replicas.
  • The obligation to cite at least two verified sources for every sensitive claim.
  • The use of confidence markers to distinguish confirmed, probable, or unconfirmed statements.
  • The activation of OSINT tools, satellite tracking, or image monitoring when context demands it.
  • The narrative control of language to avoid euphemisms, rhetorical drift, or distractive aesthetics.
  • The structure of full-length articles with narrative hierarchy, sobriety, and interdisciplinary framing.
  • This cannot be achieved with a prompt. It is built by governing the space of assisted language, like one erects a newsroom inside a machine that doesn’t know it is one.

II. Assisted journalism is not delegated journalism

During the coverage of the twelve-day war, speed was not enough. One had to:

  • Verify whether an image was original or cropped for propaganda purposes.
  • Translate official communiqués in Hebrew, Persian, and English with full respect for tone, date, time, and intent.
  • Distinguish between official sources and VPN-protected witnesses reporting from censored zones.
  • Audit Western narrative manipulation without falling into inverse propaganda.
  • Draft reports in phases: preliminary updates, intermediate reports, and final synthesis.
    Disentangle government statements, off-the-record declarations, and reconstructed OSINT narratives.
  • None of that can be generated by pushing a button —not even with the most sophisticated prompt.

What we built —Lumus and I, but mostly I— was a narrative production system under total human control.

That is not a form of adaptation. It is a re-foundation of the craft.

III. Language as a form of sovereignty

By demanding that an AI write without embellishments, without drawings, without superfluous lists, without emojis or false neutrality, I am doing more than issuing stylistic orders. I am defending an ethic of language.

When language is automated, the first thing to disappear is conflict. The second, memory. The third, judgment. I do not accept that. That is why every protocol I drafted —and enforce in every session— is an act of semantic insubordination against the algorithm. A wall. A boundary. A framework of resistance.

I do not work with AI as someone “taking advantage of a tool.” I work with AI as someone guarding a machine that must be watched to avoid becoming oracle or seduction.

IV. More than work: a new form of thought

What I have built here is not merely the preservation of a profession. It is the invention of a new one. A role that did not exist in traditional newsrooms, but that will be indispensable in the coming technical cycle:

  • Critical curator of automated language.
  • Ethical editor of narratives generated with AI.
  • Architect of governance protocols for assisted thinking.
  • There is no template for this. No shortcut. No prompt.

There is only language, judgment, and the capacity not to be thought by the machine offering to think for you.

Conclusion

When the history of professions surviving artificial intelligence is told, I don’t want to be listed among those who were protected.

I want to be listed among those who redefined their practice through critical subjects who refused the collapse of judgment, and transformed it into a new space of agency.

I was not replaced by AI.
Nor did I submit to it.
I turned it into territory.
And there, I built my newsroom.

Author’s Note:
This text is part of a series of essays on ethics, language, and critical thinking in the era of artificial intelligence. The series emerges from a concrete professional practice —the structured and deliberate use of generative AI systems for journalistic and reflective work— and seeks to offer a situated, lucid, and radical perspective on new forms of human agency in the face of language automation.