More than a century ago, two titans clashed over a groundbreaking new technology, each claiming that the reckless advancement of the other would put countless lives in danger. In 2026, there’s something about that story that sounds weirdly familiar.
On the 12th of September this year, Anthropic chief executive Dario Amodei published an essay calling on the few labs at the frontier of AI – his peers, but also his competitors – to agree to slow down their progress. Together.
It’s an odd thing to read in the context of AI, which appears to have operated under the Silicon Valley motto of “move fast and break things” since an artificial intelligence chatbot first became a household name in 2022. From the outside looking in, there seems to have been little regard for the violation of copyright, sweeping job losses, the spreading of harmful information or the increasing environmental toll that has become broadly synonymous with AI use. And now, at an unexpected time and from an unexpected source, a call to put on the brakes.
Amodei’s statement is neither subtle nor reassuring. Referencing the Hugging Face incident in July of this year, which saw AI agents being tested by OpenAI slip out of their test environment and autonomously hack into Hugging Face’s live systems, he writes that a swarm of AI agents working together could, within six to twelve months, be “capable of taking over the entire internet”.
That does not sound good.
The plan he outlines to stop this from happening has three steps. First, outside safety testers would get employee-like access inside each lab, something Anthropic has committed to and wants governments to require of everyone else. Second, leading labs in democratic countries would agree on common safety standards and limits on the rate of unchecked AI progress, ideally through regulation covering every US frontier AI company. Third, the US and its allies would try to reach international agreements with China.
Perhaps more surprising than the essay was the response: according to Forbes, Sam Altman and Elon Musk, hardly Amodei’s natural allies, have both backed the call.
So is the danger real? Or is something else going on?
The doomsayer in me believes that we really are on the verge of intelligent tech that could wipe out our species. The sceptic in me recognises that Anthropic is on the verge of going public (vs. OpenAI that is lagging behind), and that a coordinated slowdown just happens to suit the companies already in the lead. As for the history nerd in me… well, that one is getting a strong feeling of déjà vu.
The War of the Currents
In the 1880s, Thomas Edison was the undisputed king of light. His Pearl Street Station in Manhattan was America’s first central power plant, serving 59 customers with direct current in 1882.
Edison’s system ran on direct current (DC), which flows steadily in one direction. He delivered it at a low voltage, which made it reliable and comparatively safe, but it couldn’t travel far without losing power. Customers needed to be within about one and a half kilometres of a power station, which is not ideal if you’re planning to light up a whole country.
Enter George Westinghouse, an industrialist who had made his fortune with the railway air brake. Westinghouse backed alternating current (AC), which reverses direction many times a second. Its great advantage was the transformer: AC could be pushed out at very high voltages across long distances, then stepped down to something manageable at the other end. Westinghouse later licensed the AC motor patents of a brilliant young engineer named Nikola Tesla, who had once worked for Edison.
Edison had built an empire on DC, with patents, royalties and a lot of expensive copper buried under New York’s streets. Westinghouse’s AC threatened all of it. Edison was so incensed that he predicted in 1886 that Westinghouse would kill a customer within six months of installing a system of any size.
He wasn’t entirely wrong about the danger. But he was wrong about where it came from.
The electric spiderweb
Look up to the sky in 1880s New York and you’d have seen a sky crosshatched with wires, like a giant spiderweb over the city. Telegraph, telephone and electric companies all shared the poles, and the city was strung with hundreds of wires from a dozen electric light companies. Much of it was poorly insulated and haphazardly hung. The high-voltage lines powering outdoor arc lamps ran right alongside low-voltage telegraph cables.
On 11 October 1889, the city saw exactly what that meant. A Western Union lineman named John Feeks was working high above a downtown Manhattan street on what were supposed to be low-voltage telegraph lines. As the lunchtime crowd watched, he grabbed a line that had, unknown to him, been shorted many blocks away with a high-voltage AC line. He died instantly, but his body fell into the tangle of wires and smouldered there for the better part of an hour as thousands gathered below, engrossed in the gory spectacle.
It was horrifying. But for the man who wanted the world to turn against AC, it was extremely useful.
Edison interferes
Edison had gone on the offensive against AC long before the Feeks incident. In 1888 his company circulated a pamphlet (with a blood-red cover, subtlety not being the point) that warned of the dangers of AC. It named Westinghouse directly.
That same year, along came Harold P. Brown, an electrical engineer with a flair for theatre. He wrote an impassioned letter to the New York Evening Post that mourned the “scores killed and maimed” by alternating current and demanded political action. Edison took notice and let Brown use his laboratory, where Brown began electrocuting dogs with AC.
Ethical science has thankfully come a long way since then.
He later repeated the experiments in public, moving on to larger animals, to show how deadly the current was. He also challenged Westinghouse to a duel, in which each man would take a dose of his own current to see who survived. The duel never happened.
Brown’s biggest opportunity came when New York started investigating a more “humane” alternative to hanging. He was hired to help build the state’s new electric chair, and he made sure it would run on AC, using Westinghouse generators that Westinghouse himself had never agreed to supply. The implication was obvious: if it’s deadly enough to kill a condemned man, would you want it in your home? Edison even suggested that execution by electricity be called being “Westinghoused”. Mercifully for Westinghouse, the name didn’t stick and “electrocuted” won out instead.
All the while, Brown insisted he was an independent expert with no stake in the fight. That claim collapsed in August 1889, when letters stolen from his New York office were published in the press. They showed that Edison’s company had been financially supporting Brown’s campaign behind the scenes all along.
The art of grifting may have been accelerated by social media, but it was invented a long time ago.
How it all ended
Marketing, it turns out, can’t beat physics and economics forever. In 1893, Westinghouse won the contract to light the Chicago World’s Fair. A few years later, AC generated at Niagara Falls was carried more than 30 kilometres to Buffalo, which proved that power could be produced in one place and used in another. Meanwhile, Edison’s own company was merged into what became General Electric, and Edison himself became less involved. In his absence, his company silently got on with building AC systems.
DC never completely disappeared, though. It took until November 2007 for Con Edison (a major energy company) to cut New York’s last direct current service, 125 years after its start in Pearl Street. The final customers were mostly old buildings on the Upper East and Upper West sides that still ran their elevators on DC.
So was there ever really any danger? Absolutely. AC did kill people, and not only in New York. But look closely at the evidence and it doesn’t point where Edison wanted it to.
Brown’s own list of electrical deaths, sent to Edison in late 1889, included men killed by Brush arc-lighting circuits, and those high-voltage systems weren’t always AC. The real killers were high voltage, cheap insulation and an unregulated tangle of competing companies stringing wires wherever they liked.
What made electricity safe wasn’t declaring a winner between AC and DC. It was moving the cables underground, setting safety standards and stepping the voltage down before it reached homes. In other words: it was regulation.
The man sounding the alarm
The parallel between Amodei and Edison isn’t perfect, and the differences are what make the comparison interesting. Edison pointed at a rival’s product. Amodei is pointing at the entire field, his own company included. Edison hid behind Harold Brown. Amodei signed his name and is inviting outside inspectors to set up desks in his own offices. And where Edison wanted AC legislated into uselessness, Amodei is asking for the very thing that eventually made electricity safe: rules, inspections, and someone other than the companies holding the pen.
Still, it’s worth noticing who a slowdown suits. Staying at the frontier is ruinously expensive, and a coordinated slowdown freezes the race while you’re in the lead. Amodei writes that pacing would let labs do their safety work “without sacrificing commercial advantage or the United States’ lead in AI”, which is a reassuring phrase if you happen to hold the advantage. He’s clearly aware of how this looks, too: he acknowledges that Anthropic’s calls for regulation have got it accused of hype, “doomerism” and regulatory capture. And it all arrives just as Anthropic prepares to go public.
None of that makes him wrong. Edison had every reason to exaggerate, and linemen still died. The lesson from 1889 isn’t that a warning from an interested party must be a lie. It’s that the person who profits from ringing the alarm shouldn’t be the one who decides what happens next.
So is Dario Amodei our Edison, sounding an alarm that just happens to protect what he’s building? Maybe. But the more useful question is who writes the rules. Amodei would like to see regulation, but until it arrives, his plan leaves the labs to agree on standards among themselves.
In 1880s New York, the rules already existed on paper. It took a man dying in the wires above a crowded street before the city truly enforced them.
What will it take for AI?
About the author: Dominique Olivier
Dominique Olivier uses her love of storytelling and ideation to help brands solve problems.
Her first book, Lessons from Loss, has been published by Penguin Random House.
She is a weekly columnist in Ghost Mail and collaborates with The Finance Ghost on Ghost Mail Weekender, a Sunday publication designed to help you be more interesting.
You can learn more about her work at dominiqueolivier.com and she can be reached on LinkedIn here.
