The Manufactured Eschatology
A Cal Bay AI℠ Essay
Introduction
Eschatology is the study of the end of the world — the last days, the final judgment. In recent years, much of the public conversation about artificial intelligence has taken on exactly that tone: machines that copy themselves, engineered pathogens, runaway cyberweapons, and the end of human agency.
These warnings are usually presented as courageous whistleblowing. This essay argues that, whatever the intentions of the people sounding them, the fear also functions as a classic industrial strategy. Wrapping commercial software in the language of existential threat helps the largest companies build regulatory walls, pressure open competition, and keep public attention away from costs that are already landing on ordinary households.
The apocalypse is in the headlines. The bill is in the mailbox.
Part I: Two Stages, One Production
The fear operates on two stages at once. One is visible to the public. The other is where the money moves.
| The Public Theater | The Commercial Bottom Line |
|---|---|
| ”Digital godhead” and extinction rhetoric | Licensing and reporting rules tied to compute thresholds, such as 10²⁶ operations |
| High-profile summits and moral panic | Pressure on open-weight models that anyone can download and run |
| Debate about speculative catastrophes | Silence about rising power bills, water use, and tax breaks |
| Citizens cast as helpless spectators | Hundreds of billions in infrastructure spending protected, and pricing power over businesses that rent AI by the month |
Existential Fear → Rules Only Giants Can Meet → Fewer Competitors → Rented AI → Costs Passed to the Public
Part II: The Three Drivers of the Fear Architecture
1. The “Safety” Moat — Regulatory Capture
The central mechanism is legislative. If lawmakers come to see models trained above a certain amount of computing power as potential weapons, they write rules around that line. Policies including a 2023 U.S. executive order (since rescinded) and California’s 2025 frontier AI law have used a threshold of 10²⁶ training operations to define the models that face extra reporting and safety obligations.
| Who | What Compliance Means |
|---|---|
| The Largest Companies | Dedicated legal teams, compliance staff, and audit budgets. Extra rules are a cost of doing business — and a barrier to anyone trying to catch up. |
| Startups and Independent Developers | Licensing, audits, and testing mandates can cost more than the entire company is worth. |
In economics this pattern has a name: regulatory capture, described by economist George Stigler — when regulation ends up serving the industry it was meant to restrain. The more catastrophic the threat sounds, the easier it becomes to justify rules that only a handful of corporations can afford to follow.
2. The Open-Weight Threat
Open-weight models — such as DeepSeek, Qwen, and Llama — can be downloaded and run on anyone’s own hardware. That makes them a direct threat to a business model built on renting AI by the token.
If model weights come to be treated like hazardous materials, sharing them freely with developers, researchers, and small businesses can be restricted. This is not only theoretical. In 2025, Senator Josh Hawley introduced a bill that would have imposed prison time and heavy fines on Americans who imported Chinese AI technology — language broad enough that critics warned it could cover simply downloading a model like DeepSeek. The bill did not become law, but it showed how quickly “security” framing can reach the act of downloading software.
The picture is not one-sided. Meta built Llama as an open-weight family, and other large companies have released open models of their own. The incumbents do not all move together. But the general pattern holds: the scarier the story about model weights, the stronger the case for keeping them behind a meter.
3. The Screen for Physical Extraction
While the public debates speculative digital catastrophes, very real costs are being socialized in the physical economy. As The Structural Bifurcation describes, Main Street underwrites the power backbone of the AI boom.
| Cost | What’s Happening |
|---|---|
| Electricity Bills | In the PJM grid, which serves 13 states and Washington, D.C., projected data center demand helped drive capacity prices up roughly tenfold in a single auction. Households in data center hubs are seeing the increases on their monthly bills. |
| Water | Many data centers use large volumes of water for cooling, often in regions already under water stress. |
| Tax Breaks | Local governments grant property and sales tax abatements to warehouse-scale campuses that, once built, employ relatively few permanent workers. |
| Capital Scale | The largest technology companies have planned roughly 650–725 billion in capital spending for 2026, much of it for AI infrastructure — investment that depends on continued public tolerance of these costs. |
A public focused on the end of the world is not looking at its utility bill.
Part III: The Citizen as Spectator
The deepest effect of the fear campaign may be psychological. Apocalyptic framing casts ordinary people as helpless subjects awaiting the decisions of a few laboratories and governments. The only roles left are panic or prayer.
This is the same imported hopelessness described in The Truth Hurts First: a population convinced it has no power does not organize, does not build, and does not ask who is paying the bill. The Core Deduction calls it the digital sedative — distraction that keeps the extractive machinery insulated from scrutiny.
| What Fear Tells People | What Is Actually True |
|---|---|
| ”Only experts can understand this.” | The core mechanics can be explained in plain language, as Demystifying the Oracle shows. |
| ”Only governments and giant labs can manage the risk.” | Most real failures are engineering failures that ordinary builders can guard against. |
| ”You must rent this from someone bigger than you.” | Open models can run on modest local hardware for many tasks. |
| ”Your role is to watch.” | Your role is to understand, build, and hold institutions accountable. |
Part IV: The Strategic Realignment — Evidence Over Panic
An evidence-based approach to AI risk does not try to regulate an imagined digital consciousness. It focuses on what actually goes wrong in real systems, and on the harms that are happening now.
| Speculative Framing | Concrete Risk Management |
|---|---|
| Regulate “dangerous minds” by model size | Regulate high-stakes uses: lending, housing, insurance, hiring, medicine |
| Trust voluntary corporate pledges | Require deterministic harnesses: strict output validation, isolated runtimes, hard permission limits |
| Debate whether a model “wants” something | Verify intermediate reasoning and log every automated decision |
| Ask the public to fear | Ask companies to disclose energy use, water use, and who pays for grid upgrades |
| Concentrate power in a few licensed labs | Protect open research and local ownership |
The engineering details are laid out in Demystifying the Oracle: constrain outputs, sandbox agents, and keep a human circuit breaker on every high-impact decision. That is how real software failures are prevented — not by summits, and not by fear.
A Note on the Other Side
To be fair, many researchers — including some with no financial stake in the largest companies — hold serious safety concerns sincerely, and some risks, such as AI-assisted misuse, are taken seriously by independent experts as well. This essay does not argue that every warning is false. It argues that fear is also a commercial and political instrument, and that the public should ask, every time it hears a warning: Who benefits from this framing, and who pays for it?
Conclusion
The end-of-the-world story about artificial intelligence is gripping. It is also convenient. It concentrates power in the hands of the few companies large enough to meet the rules it inspires. It casts suspicion on the open tools that could let communities run their own systems. And it draws the eye away from the electricity bills, water tables, and tax breaks that are paying for the buildout today.
Real risk management does not regulate an imagined machine consciousness. It regulates real uses, enforces real engineering limits, and makes the real costs visible.
The way out of manufactured panic is the same as the way out of every other form of indoctrination in this archive: understand the machinery, follow the money, and build what answers to you.