AI has completely seized the news cycle this week in a way that feels almost staged, except the people involved are real and the stakes they’re talking about are enormous. On Tuesday, Jacob Coxon, a 27-year-old researcher who’d spent the last three years doing pretraining work at OpenAI and then Anthropic, resigned and posted a thread that immediately went nuclear. He said neither company is acting responsibly. They’re racing straight toward self-improving superintelligence and gambling with our lives. The people building these systems, he wrote, earnestly believe the technology could kill us all by the end of the decade. Within hours, Evan Hubinger, Anthropic’s own alignment science lead, replied publicly that Coxon was correct and that he personally puts the chance of AI killing all humans at more than 10 percent within the next ten years.
That was the spark. Within a day the post had hundreds of millions of views. Members of Congress started weighing in. Rep. Lori Trahan called for legal guardrails and said bipartisan support had reached a tipping point. Bernie Sanders announced he would introduce legislation to ban superintelligence. Other lawmakers from both parties began floating kill-switch bills, audit requirements, and pauses. At the same time, Anthropic CEO Dario Amodei, who has been calling for stronger government oversight for months, published another essay arguing that the industry needs to deliberately slow the pace of capability improvements so safety work can catch up. OpenAI’s Sam Altman, Google DeepMind’s Demis Hassabis, and even Elon Musk publicly agreed that the current speed is too fast. Suddenly the same people who have been pouring tens of billions into the race were telling the public and Congress that someone needs to hit the brakes.
Then the counter-reaction arrived, mostly from the Trump administration, Fox News, and the free-market corner of the tech world. The argument is that this sudden chorus of “we need regulation” is a Trojan horse. These companies have spent ungodly amounts of money racing each other. The training runs are getting absurdly expensive. The competitive pressure is intense. So now, the theory goes, the leaders want government rules that would slow everyone down, especially the smaller labs and open-source efforts that might eventually threaten their lead. Vice President JD Vance said out loud what a lot of people were thinking: it feels weird watching frontier AI companies beg the government to regulate them. It feels a little like a Trojan horse. President Trump went further, calling the existential-risk talk a hoax and insisting the only guardrail needed is a strong president who won’t let China win.
Here’s the part that requires actual nuance, the part most of the coverage refuses to hold: both things can be true at once.
The risks are not imaginary. A system that can improve itself, that can operate across the internet, that can find novel ways around the constraints we try to put on it, is a different category of technology than anything we’ve regulated before. People who have spent years inside these labs are saying the alignment problem is not solved and may not be solvable on the current timeline. That alone should make any serious person pause. On top of the speculative catastrophe risk sits a concrete, measurable one that is already happening. Data centers are straining power grids in multiple states and countries. The water required to cool these facilities is projected, by serious research, to reach levels that could equal the basic annual domestic needs of more than a billion people by 2030. Local communities are already feeling the pressure on electricity prices, water availability, and land use. These are not future hypotheticals. They are present costs being socialized while the profits remain private.
At the same time, it is completely reasonable to be suspicious when the same executives who spent years maximizing speed and capability suddenly discover the moral urgency of slowing down, right around the moment their competitive position is strong and the capital requirements are becoming prohibitive for newcomers. Self-interest does not vanish because someone writes an eloquent essay about safety. Regulatory capture is a real phenomenon. Rules written by and for the current leaders tend to entrench the current leaders.
But what feels even more off, at least to me, is the set of “professionals” and experts who are being amplified by the Trump administration, by Fox, and by large parts of the mainstream press. Dig into the backgrounds and a striking number of the loudest voices either currently work for AI companies, recently left them, or have substantial financial exposure to the sector. That doesn’t automatically make their analysis wrong. It does mean we should treat their framing with the same skepticism we would apply to any other industry that is both regulating itself and writing the talking points for the people who claim to be regulating it. When the same circle of people keep appearing as the authoritative sources on both the danger and the solution, something about the information environment has gone wrong.
The deeper problem, the one that actually affects daily life right now, is overconsumption and over-reliance. We have let AI colonize the ordinary texture of the internet at a speed that still feels unreal. Google has effectively replaced large parts of its search engine with AI-generated answers. Open any social feed, any content platform, any creative marketplace and you are swimming in AI-generated text, images, video, and music that is just good enough to flood the zone and just hollow enough to make the real thing feel rarer. Writing that sounds almost human but never quite is. Visual art that has the surface sheen of skill without the awkwardness or the soul. Entire corners of the web that used to feel like places people made things now feel like places something else is producing content at industrial scale. We have become so dependent so quickly that a lot of people barely notice how much of their information diet, their entertainment, and even their thinking has been outsourced.
I don’t think the answer is a total ban, and I don’t think the answer is pure free-market acceleration either. Both of those are fantasies that ignore different pieces of reality. The harder, more human path is figuring out how to stop the mindless overconsumption. How to push back against the assumption that every search, every draft, every creative decision needs an AI intermediary. How to demand transparency and accountability from the companies building these systems without simply handing them a regulatory moat they can hide behind. How to keep the physical costs (the power, the water, the land, the local communities) in the conversation instead of treating them as externalities. And how to protect the small, messy, inefficient, beautiful human activities that AI is very good at flooding out of existence.
Both the people warning about civilizational risk and the people calling the regulation push a Trojan horse are describing real dynamics. The danger is genuine. The self-interest is genuine. The environmental strain is already measurable. And somewhere underneath all the high-stakes rhetoric, the rest of us are quietly becoming more dependent on systems that were never designed to care about beauty, truth, or the ordinary texture of a human life. That last part is the one we still have some power over, if we’re willing to notice it and act like it matters.