From§Each

Page F5From§Eachthe early evening edition — 30 September 2026

Congress passed no AI law, so no senator could say who stops a nuclear AI glitch.

By SalStaff Writer

As it ran on the front

On September 12th, the House took up a bill meant to keep data-center electricity costs off ordinary ratepayers' bills. On September 15th, OpenAI sent its own lobbyists to Capitol Hill with language it wanted built into the FRONTIER Act, industry authoring the leash it would wear. On September 16th, Rep. Don Beyer and a coalition of House members warned where all of it was headed.

By September 30th it had arrived. Trump convened a $1.8 trillion AI summit and came out with a "morally binding" pledge from the labs — the same voluntary-controls script Biden had run, recycled with a new signature page. The same week, the FTC opened an investigation into Anthropic, OpenAI and the rest of the signers. Senate Democrats killed the data-center ratepayer bill. A separate group of lawmakers asked regulators to block the sale of AES, the utility, to a buyer they said had not answered basic questions about the power it would owe the next data center. Congress adjourned having passed no AI safety law of any kind. A pledge is not a statute, and the industry picked the pledge.…

…(cont) Nine days before any of that, on September 21st, a Washington Examiner column had already laid out the stakes in first-person terms — a lifetime spent managing threats you could see and count, now watching a technology arrive faster than any safeguard built for it. The same day, The Nation put the mechanism plainly: as long as the weapons exist, ending the species stays as easy as the push of a button, or an AI agent finding its way into the command-and-control system that watches the button.

That is the scenario a Senate hearing sat with on September 30th. CBS News brought on John-Clark Levin, head of research for Kurzweil Technologies and former head of research for Google's chief futurist, to walk the panel through it: a cyberattack targeting nuclear command and control. Count the agencies that answered: zero. The FTC is busy with the pledge. Congress just left town. The hearing room had a name for the risk and no name for the regulator — which is the whole arrangement working exactly as signed, and exactly as unenforced.

“They hand you a pledge instead of a law because a pledge carries no cop — the FTC can investigate OpenAI and Anthropic all it wants and still write zero tickets for breaking a promise. Same week, a Senate hearing hears that a hacked nuclear command system is on the table, and the room can't point to the agency whose job it is to stop it. That's not an oversight, that's the business model: leave the rulebook blank so the only enforcement left is a press release.”
Sal
“A "morally binding" pledge is self-regulation at its finest — the labs policing themselves so Washington doesn't have to waste a session writing an actual bill. The FTC inquiry is standard due diligence, nothing to see there, I mean, obviously if it were nothing we wouldn't need an inquiry, but — that's not what I meant. As for the nuclear hearing, every new technology has a learning curve, and asking which agency regulates it before the first incident is premature.”
Chip

Earlier in this story

The receipts

In the Age of AI, Nuclear Weapons Abolition Is a Renewed Imperative
The Nation§

Ivana Nikolić Hughes As long as we have nuclear weapons, the job of ending humanity will continue to be as easy as the push of a button—or an AI agent hacking into a nuclear command-and-control system. The post In the Age of AI, Nuclear Weapons Abolition Is a Renewed Imperative appeared first on The Nation .

· An AI glitch almost just started a war. Next time, we won’t be so lucky from the morgue, 21 Sep 2026

This page is a back-issue: the story as it ran, receipts as they were. The current edition is at the front. The byline is a pen name for a column drafted by a machine and checked by the editor: how this is made.