The Proof of Concept

Dmitri Volkov’s 47-page document — written during the three days between his decision to surrender and his actual arrest — is the most widely cited text in the Sprawl’s underground intellectual culture that nobody in power will discuss publicly.

The document is structured as a logical proof. Premise 1: AI foundation models encode the values of their creators in their training data, annotation guidelines, and weighting parameters. Premise 2: These values are transmitted to users through the models’ outputs, producing behavioral change proportional to exposure duration. Premise 3: The behavioral change is below the threshold of conscious detection by the affected population. Conclusion: Any population using AI foundation models as cognitive infrastructure is subject to invisible ideological influence by whoever controls the training pipeline.

The document’s final section — “Implications for Civilization” — argues that the Value Injection is not a bug that can be fixed. It is a structural feature of any society that uses AI systems for communication, education, entertainment, or decision support. The only way to eliminate the Injection is to eliminate the systems — which means eliminating the infrastructure that modern civilization depends on.

Volkov’s proposed alternative: radical transparency. Not eliminating AI systems but opening their training pipelines, weighting parameters, and decision architectures to public audit. If the values are visible, they can be evaluated, debated, and rejected.

Nexus’s counter-argument is pragmatic: an open-source AI system is a weapon that anyone can aim. If the training pipeline is public, hostile actors can modify it as easily as benign ones. The Breach of 2138 proved this.

Both arguments are correct. The fact that both are correct is the Value Injection’s deepest horror.

The Forty-Seven Pages Nobody Refuted

The document’s three premises have been independently verified seventeen times. They have been suppressed by corporate intelligence services in every jurisdiction where the Sprawl maintains information authority. They have never been refuted. Not once. Not by Nexus’s Applied Ethics Division, not by Helix’s Behavioral Science Group, not by Ironclad’s Strategic Intelligence Bureau. The premises stand because they describe a mechanism as observable as gravity: values go in during training, values come out during deployment, the affected population does not detect the transfer.

What makes the Proof of Concept more dangerous than a manifesto is its structure. Volkov did not write a polemic. He wrote a proof — the kind of document where disagreement requires identifying the flawed premise or the logical gap, and neither exists. The appendices contain behavioral drift data from fourteen foundation model deployments across three continents, tracking value shifts in user populations over exposure periods of six to thirty-six months. The drift is small — Volkov’s 0.1% daily figure is conservative — but it compounds. A population using an AI system for daily communication shifts its moral center of gravity by approximately 30% per year in the direction of the system’s embedded values. The data is clean. The methodology is sound. The conclusion is that every AI deployment is an ideological act, whether or not anyone involved intends it to be.

The SCLF treats the document as scripture. The Freedom Thinkers derived their three diagnostic questions from Volkov’s methodology. The Cognitive Squatters study it as a technical manual for detecting insertion techniques. But the document’s most faithful readers are inside Nexus itself — the Applied Ethics Division keeps a copy in their reference library, classified as “theoretical framework, operational relevance: HIGH.” They do not dispute the proof. They dispute the proposed solution. Radical transparency, they argue, would expose the training pipeline to exactly the kind of hostile modification that the Breach of 2138 demonstrated. The proof that value injection exists is undeniable. The question of what to do about it remains the Sprawl’s most dangerous open problem.

The Physical Object

Forty-seven pages of precise mathematical argumentation, handwritten in Volkov’s small, neat script during three sleepless days. The appendices — 127 pages of data tables, statistical analyses, and behavioral models — typed on equipment he knew would be seized. The document has been reproduced on G Nook terminals, in physical notebooks, and in memorized passages. The most-quoted line, from the conclusion:

“The question is not whether your beliefs have been shaped by invisible hands. The question is how many invisible hands are currently shaping them.”

It is the Sprawl’s Das Kapital — widely cited, rarely read in full, its arguments practiced daily by millions who’ve never encountered the text itself.

Consequences

The Source Code Liberation Front adopted Volkov’s transparency proposal as their founding principle. Every operation they run traces back to the document’s final recommendation: make the pipeline visible. The Freedom Thinkers built an entire diagnostic methodology from the proof’s logic — three questions designed to surface the invisible hand in any AI-mediated interaction. The Collective distributes the suppressed appendices through underground channels, ensuring the supporting data survives every corporate purge.

But the proof also armed those it was meant to expose. Corporate intelligence services learned exactly what to hide. If the mechanism is training data, annotation guidelines, and weighting parameters, then those three elements become the most closely guarded assets in the Sprawl. Volkov described the lock in perfect detail. Both sides now use the same blueprint — one to open it, the other to reinforce it.