Every six months, Silicon Valley holds a seance. A fresh cadre of self-anointed high priests takes to the conference circuit, warning that digital intelligence is about to wake up, look at humanity like an annoying fungus, and switch off our oxygen. They clutch their pearls, testify before congressional subcommittees, and demand massive regulatory moats that happen to protect their incumbent balance sheets while keeping open-source code out of the hands of the plebeians.
Then Donald Trump looks at the circus, calls the existential risk narrative a hoax, and the technocracy loses its collective mind.
They call him dangerous. They call him anti-science. They argue he is selling out humanity for short-term electoral and economic calculus.
They are wrong. Not because Trump is a visionary tech philosopher, but because a stopped clock is right twice a day, and on artificial intelligence safety, his populist dismissal is infinitely more grounded in reality than the apocalyptic fairy tales whispered in the boardrooms of San Francisco.
Stop worrying about paperclip maximizers and rogue superintelligences erasing the species. That is not an engineering crisis. That is a marketing campaign designed to consolidate power. The real catastrophe is much dumber, far more pedestrian, and happening right now in plain sight.
The Doomsday Panic is a Protectionist Monopoly Racket
Let us look at the mechanics of the safety lobby. Yann LeCun, Meta's chief AI scientist, has spent years pointing out the absurdity of the extinction panic. Yet the mainstream media continues to amplify every bedwetter who claims that next Tuesday, an unsupervised neural net will figure out how to synthesize a bespoke pandemic in a garage.
This is not caution. It is regulatory capture wrapped in a priestly shroud.
When major labs lobby for federal licensing requirements, mandatory safety audits by approved agencies, and restrictions on open-source model weights, they are pulling up the ladder behind them. If you make compliance so expensive that only three trillion-dollar conglomerates can afford the legal and computational overhead, you do not prevent catastrophe. You prevent competition. You institutionalize a cartel.
I have watched enterprise executives flush millions down the drain on bespoke risk frameworks, hiring compliance consultants to grade papers for software that learns faster than any human committee can audit it. It is security theater of the highest order. They build complex governance frameworks because it feels productive to manage a phantom threat, whereas fixing actual data hygiene or internal cybersecurity requires boring, unsexy labor.
The extinction narrative assumes a magic-wand capability that does not exist. Current models do not possess intent, consciousness, or endogenous drive. They are magnificent, hyper-scaled statistical pattern matchers trained on the collective exhaust of human civilization. They complete text. They interpolate vectors. They do not wake up plotting our demise because they have nothing inside them that corresponds to an "us" or a "them."
The Real Crisis is Bureaucratic Sludge and Intellectual Laziness
If the doomsday panic is a hoax, what is the actual threat?
It is not that artificial intelligence is too smart. It is that humans are too lazy.
Look at what corporate America is doing with large language models. They are not building hyper-efficient autonomous agents that will optimize global supply chains. They are writing sterile corporate emails, generating mediocre marketing copy, and automating middle-management tasks that should have been eliminated a decade ago.
We are taking the most potent cognitive tool in human history and using it to scale mediocrity. Companies buy software licenses under the banner of digital transformation, pump millions into proprietary APIs, and then wonder why their productivity metrics remain flat. They substituted genuine thought with automated autocomplete and called it innovation.
Worse, we are breeding a generation of knowledge workers who treat the first output of a prompt as gospel. When you outsource cognitive friction to a model that tells you what you want to hear, you atrophy the exact muscle required to spot structural errors. The danger is not that machines will outsmart us. The danger is that we will happily surrender our capacity for rigorous thought because clicking a button feels like working.
The Open-Source Imperative
The entire premise of the AI safety movement rests on centralization. They want a priesthood holding the keys to the kingdom, doling out safe, sanitized intelligence to the masses through tightly controlled subscription gates.
This is historically illiterate. Centralized control of foundational technology has never yielded safety; it has only yielded tyranny, vulnerability, and stagnation.
Open-source weights—the models developed and shared by communities rather than walled gardens—are the only check against corporate and state overreach. When anyone can download a model, fine-tune it locally, and run it without a cloud connection, power is decentralized. Yes, bad actors can abuse open tools. Bad actors abuse encryption, automobiles, and the English language. But the defensive utility of open intelligence vastly outweighs the fringe risks of restriction.
If you want a resilient society, you do not lock the source code in a vault and hand the combination to three billionaires and a revolving door of government bureaucrats. You distribute it wildly, break the monopolies, and let thousands of independent developers stress-test the edges.
Stop Fighting Phantoms
The next time someone tells you we need to pause development to save humanity from an omniscient digital god, ask them who benefits from the pause. Ask them why the companies shouting loudest about safety are also the ones hoarding the most compute clusters.
The emperor has no microchips, and the doomsday prophets have no clothes. Trump calling the safety panic a hoax was dismissed as anti-intellectual populism, but beneath the rough edges, it exposed a vital truth: the real risk is not what the technology will do to us tomorrow, but what our own institutional cowardice and corporate greed are doing to our capability today.
Stop writing safety manifestos. Start auditing your actual codebases, fixing your broken data pipelines, and teaching people how to think critically instead of prompting blindly.
The machines are ready. The problem is everything sitting in the chair.
