Have you noticed how quickly a new technology stops being a tool and starts sounding like a sermon? I have. One week it is chips, models, and power bills. The next week it is prophecy. People who spent years rolling their eyes at old-time religion now talk about software as if it were a Messiah with an expiration date stamped on humanity. That shift is not a curiosity. It is becoming policy theater, and the theater has a cost.
When Code Starts Sounding Like Creed
A senior technology founder recently put the mood in plain language. The loudest alarm, he argued, is not a sudden burst of engineering caution. It looks like a coordinated effort to frighten the public into slowing the one sector that still gives the United States a hard edge. I am not claiming every safety researcher is acting in bad faith. Plenty of them are doing real work on real failure modes. What I am saying is simpler. Fear, once packaged, travels faster than a white paper.
The founder’s framing was blunt. Many of the most vocal doomers no longer hold a traditional faith. They still want a big story. They want a messiah and an ending. Artificial intelligence became both. Call it a machine. Call it a model. In that circle it functions like a sacred object. Treat it as holy and you can demand that the state kneel with you.
These guys don’t believe in God. They’re atheists, but they’ve created something they believe is God.
That line is easy to mock. It is also hard to ignore once you listen to the language. End of the world. Gambling with our lives. Alien intelligence. Those are not lab notes. Those are revival-tent words wearing a hoodie. I’ve found that when a debate swaps measurement for mythology, the next request is almost always the same: more power for a referee, less room for builders.
The San Francisco Vacuum And The Search For A Messiah
One coastal culture dropped a lot of inherited religion and did not become calmly secular. It became hungry. If you give up an old faith, you still want a large thing to believe in. Effective-altruist circles, in this telling, filled the hole with an industrial-scale obsession. The model is the Messiah. The chart is the apocalypse. The obsession is the liturgy.
Does that mean every researcher in those rooms is a zealot? Of course not. Some of the best safety work comes from people who simply hate sloppy deployment. The problem is the public script. The script does not ask for narrow rules on the handful of systems that can actually do damage. It asks for a mood. Moods become statutes. Statutes become moats.
In my experience, industries do not get regulated in the abstract. They get regulated after a scare has already chosen winners. The scare does not have to be fake to be useful. It only has to be loud enough that lawmakers feel they must “do something” before they understand what they are doing.
Scare First, Then Write The Rules
The sequence is familiar if you have watched other issue cycles. Announce a cliff. Recruit cameras. Demand a pause that just happens to favor the largest labs and the most political operators. Responsible guardrails are one thing. A stampede is another.
After a researcher left a frontier lab and warned that teams were gambling with lives, the same founder answered with a cooler tone. The world, he said, is going to be alright. Leaders have hard jobs. Scaring everyone does not help. That is not a claim that risk is zero. It is a claim that panic is a poor design tool.
- Real engineering limits on tools that can enable bioterror or mass-scale hacking
- Export controls that keep the most sensitive systems away from a rival state
- Clear liability when a product is shipped with known, reckless gaps
- A public mood that treats every model as an approaching deity
The first three can be argued in daylight. The fourth is a church with better Wi-Fi. Once the fourth wins the room, the first three get buried under slogans.
The Same Apocalyptic Crowd, A New Object
A political commentator spotted the pattern immediately. Climate. A public-health emergency. Now silicon. Always the same tone. Always the same demand that ordinary people accept a grand pause designed by people who will not live with the downside of losing the race.
Responsible guardrails are one thing, but handicapping American innovation while a strategic rival speeds ahead with almost no matching restraint is not caution. It is a gift.
I do not buy the idea that every prior scare was invented from nothing. Some risks were real. Some still are. The repeating feature is the personnel and the style. Catastrophe as identity. Inspection as virtue. Delay as moral high ground. If your self-image depends on being the adult who saw the end coming, you will not celebrate a boring, successful deployment. You will hunt the next chart.
Perhaps the most interesting aspect is how quickly “safety” becomes a foreign-policy question. If American labs slow down while another government treats the same technology as a weapon and a party tool, you have not made the world safer. You have changed who holds the stack.
Beijing Is Not Hosting A Pause-And-Pray Summit
Chinese firms have been accused by U.S. officials of stripping American models at an industrial scale. That charge matters whether you love open weights or hate them. A rival that can copy, distill, and redeploy does not need your permission to keep building. It needs your hesitation.
A senior security official in that system has described artificial intelligence as a threat to political security, institutional security, and ideological security. Read that twice. The concern is not a shared conscience. The concern is control. The party wants the capability and the narrative. It does not want a transnational inspector wandering through its labs with a clipboard from Davos.
So the practical question is ugly and simple. If one side writes sermons and the other side writes procurement orders, who finishes first? I am not a hawk for sport. I just do not like strategy that assumes the other player will adopt your guilt.
Washington’s Split Screen
The current administration’s public line has been impatient with slowdown talk. Lead or lose. Whoever wins this stack wins a larger contest. Horror stories, in that view, are being shopped. Some will not happen. A “sick conspiracy” against data centers, one message suggested, has an obvious beneficiary across the ocean.
Treasury leadership made a related point when lawmakers tried to treat celebrity briefings as holy writ. The people touring the panic could stop building any time they want. They have not. They want Congress to stop it for them. That is a revealing preference. Self-restraint is available. Political restraint is more useful if your goal is to lock a market.
None of this requires you to trust every executive on the growth side. Companies hype. Companies hide. Companies love a rule that raises a rival’s cost more than their own. A grown-up policy would notice that and still refuse to confuse a press tour with a constitution.
The Global Watchdog Pitch
This week the political demand behind the theology arrived in familiar packaging. A famous philanthropist and software billionaire returned to the emergency circuit. He called the technology an “alien intelligence” no government is ready for. The proposed answer was not a targeted statute. It was an international organization to police the field.
Governments, he said, are way behind. Industry should not self-regulate or even claim to understand the whole-of-society impact. He reached for the movie script in which rival powers magically unite because something from space has landed. Then he said the new systems are kind of like that. Better do what the films show.
On another recording he went further. Designing the transition cannot be left to markets, he warned, because the systems will be built by and for the richest people on earth. The good outcome will not happen by accident. In a longer essay he called the period one of the most turbulent in human history and floated a permanent cross-border body stitched together from nuclear inspections, aviation rules, and old environmental treaties. He has pitched versions of the idea to the current White House team and says he wants a meeting with Beijing’s leader later this year.
His foundation then pledged a large sum over two years to push the same technology through education, health, and agriculture. That detail is not a smear. It is a reminder. The person selling urgency is also buying pipeline. Panic and patronage can share a calendar.
Left only to the market, the argument goes, the stack will serve the rich. Left to a new global office, it will serve whoever staffs the office.
What An International Referee Actually Does
Inspectors sound neutral until you ask who writes the checklist. Aviation rules work because planes fall out of the sky in public. Nuclear inspections work, when they work, because the physics is constrained and the club is small. Software is copyable. Weights leak. Talent moves. A lab in one country can distill a model trained in another. A treaty written for reactors does not map cleanly onto a git repo.
There is also the small matter of enforcement. A watchdog that can wander through American facilities while a closed political system smiles for the photo is not symmetry. It is a one-way mirror. I’ve watched too many “shared frameworks” become lectures aimed at open societies and shrugs aimed at closed ones.
- Define a narrow class of truly dangerous capabilities instead of baptizing every chatbot.
- Keep export and procurement rules tight where a rival state is the customer.
- Fund evaluations that can be reproduced, not sermons that cannot.
- Refuse a pause designed abroad and enforced mainly at home.
That list will bore people who want a movie ending. Good. Boredom is often the adult temperature.
Politicians Smell A Stage
A former president told his party to get on top of a technology moving very fast in private hands. A senator called a runaway system humanity’s problem and again demanded a pause-and-bargain at a coming leaders’ meeting. House members rushed to agree. Entertainment circuits are already warming the next documentary tone. None of that is a safety protocol by itself. It is a bid to internationalize control of a stack American labs still lead.
Why does that matter for readers who do not live in policy group chats? Because the cost does not stay in those chats. Compute policy becomes energy policy. Energy policy becomes permitting. Permitting becomes whether a region gets a data center or a press release about caution. Jobs follow racks. Tax bases follow jobs. If you kneecap build-out in the name of a global conscience, you still need someone to generate the power and train the next model. That someone may not share your talking points.
Narrow Rules Versus A Global Pause
There is a grown-up version of this argument and a theatrical one. The grown-up version admits that some systems should not be sold like a consumer app. Open-weight models that can automate intrusion at scale deserve a harder look. Tools that lower the skill bar for biological misuse deserve a harder look. Shipping frontier capabilities to a party-state that names ideology as a security object is not “openness.” It is naivete with a press kit.
The theatrical version wants a pause designed in salon rooms and enforced by roaming inspectors. One protects a country. The other flatters a class. I will take the first even when it is inconvenient for a lab I like. I will not take the second because it photographs well.
| Approach | What it targets | Likely effect |
| Narrow engineering rules | Specific high-risk uses | Friction where it matters |
| Export and access limits | Rival-state acquisition | Slows leakage, not research itself |
| Global pause-and-inspect | The entire frontier | Hits open builders first |
| Mythic “alien” framing | Public emotion | Rushes statutes, picks winners |
Tables flatten arguments, I know. Still, the contrast helps. If your cousin works in a mid-size lab, a global pause is not a philosophy seminar. It is a hiring freeze with better vocabulary.
Is The Industrial Revolution Real Or Just A Pitch?
The same founder who mocked the messiah complex also said the technology is real. An industrial shift is coming. It could be amazing for the country if the country gets it right. That pairing is the part the doomer class never wants discussed. Not safety as engineering. Not limited rules. A substitute church, complete with prophets, end-times charts, and a demand that government enforce the creed.
Media surrounding the debate often feeds on fantasies from science-fiction culture. Fine. Fiction is a useful warning system. It is a terrible statute. Planes did not become safe because novelists banned wings. They became safer because engineers measured failure and airlines paid for it.
Will some horror stories fail to arrive? Almost certainly. Will new failure modes appear that nobody put in a keynote? Also yes. The honest posture is dual. Build like the upside is large. Test like the downside is not a joke. Do not outsource both jobs to a choir.
Faith, Atheism, And The Need For A Big Story
I am not here to baptize anyone. Belief is not my beat. The sociological claim still deserves airtime. A class that discarded inherited religion did not become modest. It became intense. Intensity needs an object. A model that writes, codes, and plans is a tempting object. It looks like agency without a soul, which is catnip if you already think history is a puzzle you were born to solve.
Once the object is sacred, disagreement feels like heresy. That is how you get calls to “stop” as if stopping were a moral solvent. It is also how you get contempt for voters who care about electricity prices and factory floors more than alignment blogs. Those voters are not stupid. They can smell when a lecture is also a power grab.
Rhetorical question, and I mean it: if the machine is God, who appointed the priests? If the answer is “people with the best slides,” you already know how this ends.
What “Getting It Right” Could Look Like
Getting it right is not a slogan. It is a pile of unglamorous choices. Power plants that actually get permitted. Grids that do not pretend megawatts appear from vibes. Immigration rules that admit scarce talent without turning labs into leak festivals. Procurement that buys American capability instead of essays about feelings. Schools that treat computation as a trade, not a superstition.
It also means telling the public the truth about trade-offs. Faster models will break some jobs and create others. Some of the new work will be dull. Some of the old work will vanish faster than a press secretary can invent a transition fund. Adults can hear that. What they cannot hear, forever, is that the only moral path is to freeze in place while a rival treats freeze as an opening.
A rough balance worth arguing over: 40% build and deploy under daylight 30% test and red-team the dangerous edges 30% keep the stack out of a rival’s command economy
Those numbers are not science. They are a way to refuse a single-note sermon. If your only note is pause, you are not balancing anything. You are conducting.
Follow The Incentive, Not Just The Adjective
Watch who benefits if the United States slows on purpose. Incumbent labs that can afford compliance departments. Political operators who need a new emergency after the last one cooled. Foreign systems that copy first and regulate later, if they regulate at all. Watch who loses. Startups that cannot hire a theology department. Regions betting on compute as the next industrial base. Workers whose bargaining power rises when productivity rises.
Adjectives are cheap. Safe, aligned, responsible, global. Incentives are not cheap. If a rule raises your rival’s cost more than yours, you will discover a sudden love of responsibility. That does not make the rule wrong. It makes the press conference incomplete.
I’ve found that readers can handle incompleteness better than elites think. What they resent is being told that skepticism is cruelty. Skepticism is how you keep a church from annexing a statute book.
The China Variable Without Cartoon Villains
It is possible to take the rivalry seriously without turning every sentence into a poster. The other system has engineers. It has capital. It has a political machine that can order energy and silence dissent. It also has fragilities. Talent wants options. Copying is not the same as inventing forever. Party control can choke the very curiosity that makes research compound.
None of those fragilities save you if you volunteer to go second. A race is not a morality play. It is a clock. Clocks do not pause because a panel in a glass building found the right metaphor.
U.S. officials already talk about industrial-scale stripping of models. Treat that as a fact pattern to test, not a vibe. If it holds, then “openness” without strategy is just a conveyor belt. If it does not hold, you still need a theory of leakage that is more adult than hope.
Why The Messiah Frame Is Strategically Dangerous
Sacred objects demand purity tests. Purity tests punish partial progress. Partial progress is how industries actually get safer. Seat belts did not wait for a world government of highways. Aviation did not freeze fleets until every nation signed the same hymn. You fix the wing. You fly again. You publish the failure. You fix the next wing.
If code is God, then a bug is sin and a delay is repentance. That is a terrible way to run a lab and a worse way to run a country. Labs need iteration. Countries need growth, deterrence, and enough humility to admit that no committee will see the tenth-order effect in advance.
The founder’s closing logic cuts through the incense. A class of people who no longer believe in God built a machine they discuss as if it were God, then demanded that Washington and a new global body kneel with them. Whoever wins this stack wins a larger contest. The people treating the code as their Messiah are asking one country to lose on purpose.
A Note On Tone, Because Tone Is Doing Work Here
You can hear the sneer in some of the anti-doomer talk. You can hear the tremor in some of the doomer talk. Both are human. Neither is a substitute for a rule that can be written, tested, and repealed. I would rather a slightly rude engineer who ships evaluations than a gentle keynote that ships a bureaucracy.
That is a personal preference, and I will own it. Gentle bureaucracies have a habit of arriving late to the harm and early to the veto. Rude engineers have a habit of breaking things and then, if the culture is healthy, writing down how they broke.
Culture is the sleeper issue. A lab culture that hides near-misses is a problem. A political culture that treats every near-miss as proof we must anoint a planetary viceroy is also a problem. You can walk and chew. You do not need a new Vatican of weights and biases to do it.
What Readers Should Watch Next
Watch the verbs. Pause. Ban. Inspect. Harmonize. Those verbs travel together. Watch whether proposed rules distinguish a consumer chatbot from a system that can plan intrusion or wet-lab assistance. Watch whether energy permitting speeds up or drowns in the same month leaders praise “leadership.” Watch whether the loudest moral language appears on the same week as a fundraising or a foundation pledge.
- Does the proposal name a concrete capability or only a feeling?
- Does enforcement hit open jurisdictions harder than closed ones?
- Does the speaker still ship products while asking others to stop?
- Does the China paragraph arrive as strategy or as decoration?
Those questions are not anti-safety. They are anti-fog. Fog is how a coordinated campaign works even when nobody sends a memo titled “coordinated campaign.” People can share a mood without sharing a conspiracy. Moods still move votes.
The Bottom Line Without A Choir
AI is not a cartoon demon and it is not a cartoon angel. It is a general-purpose stack arriving while two large systems compete over industry, information, and force. Treating it as a substitute religion helps a certain coastal identity. It does not help a country that still has to generate power, train talent, and deter a rival that will not attend the altar call.
Narrow rules for real dangers are compatible with speed. A global pause designed as liturgy is compatible with losing. I would rather argue about megawatts, evals, and export lists than about who gets to play high priest of the next model drop. That argument is less cinematic. It is also closer to work.
If a machine is going to change the industrial base, the grown-ups in the room should sound like builders and testers, not like a congregation waiting for the sky to tear. The sky may yet surprise us. Until then, the country that keeps its nerve and its grid will write more of the next chapter than the country that outsourced its nerve to a watchdog with a beautiful mandate and no matching adversary.
That is the frame worth keeping when the next interview arrives with the same adjectives and a new end-times slide. Ask who kneels. Ask who builds. Ask who copies. Then decide whether you want policy written by engineers with scars or by people who found a new God and want the rest of us to tithe in the form of delay.
