Programming note: Given my conflicts here, I sat out the writing for this one.
asserted
I → give → one
But I did edit it.
asserted
I → edit → it
And now, from the company that has promised to soon replace us, comes the biggest IPO in history.
asserted
IPO → promise → history
Anthropic raised $81.3 billion on Thursday in its Nasdaq debut, the largest-ever haul from an initial public offering.
asserted
Anthropic → raise → offering
At close, Anthropic was valued at $1.76 trillion, making the AI lab both the world’s largest public benefit corporation and one of its largest public companies, period.
asserted
lab → value → companies
On paper, today should feel like a big win for the AI industry: a darling of the AI boom just successfully went public, affirming the massive demand for frontier-level intelligence while returning hundreds of billions of dollars to the tech ecosystem.
asserted
darling → feel → ecosystem
Amid ongoing concerns about the AI bubble, from the industry’s circular-financing arrangements to the geopolitical consequences of Jensen Huang’s affairs, investors still proved eager to bet big on Anthropic.
asserted
investors → prove → Anthropic
On the other hand, Anthropic’s mission is to build AI systems that are smarter than humans — systems that its founders have repeatedly warned could kill everyone.
uncertain
founders → build → everyone
Which makes this moment … bittersweet?
asserted
moment → make → ?
It depends on who you ask.
asserted
you → depend → who
In a speech at the Nasdaq, Anthropic CEO Dario Amodei tried to walk his now-familiar line, talking up AI’s potential for scientific breakthroughs while also warning of its potential for misuse.
asserted
Amodei → try → misuse
“My company’s co-founders and I started Anthropic because we believe that the most powerful technology of our time should be built for the benefit of humankind,” he said.
asserted
he → start → humankind
Not everyone is buying it.
asserted
everyone → buy → it
While the founders popped (we assume) ethically sourced non-alcoholic champagne on Wall Street, more than 100 protesters gathered outside the AI company’s San Francisco headquarters holding signs saying “Stop the AI race,” “Earth is in danger,” and “Please stop.”
asserted
Earth → pop → danger
Among them was Jacob Coxon, whose viral tweet about quitting Anthropic drew fresh attention to the AI safety movement, and who spoke at the protest.
asserted
who → quit → protest
Four people were arrested for allegedly blocking the entrance to the company’s office.
uncertain
people → arrest → office
Elsewhere, a diverse cast of characters, ranging from perpetual AI skeptic Gary Marcus to wild-eyed AI accelerationist Guillaume Verdon (aka Beff Jezos), signed “The MisAnthropic Statement.”
asserted
cast → range → Statement
The statement reads:
“Anthropic warns that recursively self-improving AI could concentrate power and escape human control, yet races to build it first.
uncertain
AI → read → it
The company uses safety branding to court investors, employees, and the public — even as it has quietly weakened its safety pledges and lobbied for self-serving regulation.
asserted
it → use → regulation
Anthropic is not a safe AI company, and we must reject its reckless attempt to amass power in the name of the greater good.”
asserted
we → reject → good
Futurists have long theorized about the dangers of “recursive self-improvement,” a state in which sufficiently advanced AI builds next-generation versions of itself in ever-tighter loops.
asserted
AI → theorize → loops
The process risks spiraling out of control, given that humans will become less and less able to understand or stop the system when it misbehaves.
asserted
it → risk → system
A recent report from AI auditor METR concluded that Anthropic’s rate of AI R&D progress had roughly doubled due to AI’s research ability.
asserted
rate → conclude → ability
Notably, doubling the rate of research progress would indicate that the company had surpassed its own risk threshold for “Automated R&D” in its “Responsible Scaling Policy.”
asserted
company → double → Policy
According to that policy, Anthropic should now be taking on new safety auditing, security policies, and transparent reporting to meet its commitments.
uncertain
Anthropic → accord → commitments
Anthropic endorsed California law SB 53, which asks companies to publish safety policies that they are legally required to follow.
asserted
they → endorse → that
But after the law passed, Anthropic submitted a weaker framework under SB 53, recasting its more ambitious safety commitments as voluntary ones.
asserted
Anthropic → pass → ones
During his speech at the Nasdaq, Amodei promised to “pace the training and deployment of AI systems, taking as long as is necessary” to ensure they are safe.
asserted
they → promise → long
Amodei added, “We will not bend to either business nor political pressure.”
asserted
We → add → pressure
Anthropic will continue to face business pressure to keep building the most capable AI systems, though, both to retain customers and to appease its investors.
asserted
Anthropic → continue → investors
(The Financial Times’ Robert Armstrong estimated the company will need to generate more than $600 billion in annual revenue to return the 20 percent on invested capital that its peers do; Apple generates a mere $478 billion in revenue.
asserted
Apple → estimate → revenue
In a conversation with CNBC, Amodei said Anthropic aspires to be “investment grade” — to earn a high credit rating — and plans to issue more debt.
asserted
Anthropic → say → debt
Raising billions of dollars of capital and additional debt to acquire ever more compute strikes some people as an odd way to “pace the frontier.”
asserted
Raising → raise → frontier
Amodei has justified the company’s practices as being necessary to trigger a “race to the top,” in which Anthropic’s financial success and safety commitments inspire its competitors to adopt similar practices in hopes of catching up.
asserted
success → justify → hopes
Jasmine Sun, writing for the New York Times, reported hearing a striking version of this vision from an anonymous ex-employee:
“One former Ant referenced J.R.R. Tolkien’s parable in trying to explain Anthropic’s mind-set: Whoever builds superintelligence first will get Sauron’s ring.
asserted
builds → write → ring
That person or company will possess the most dangerous cyberweapons in the world, the cures and recipes for every disease, and an army of tireless digital geniuses to do its bidding.
asserted
person → possess → bidding
The most load-bearing question, then, is the morality of the one who wields the ring.
asserted
who → bear → ring
”
Many readers may remember that the One Ring from The Lord of the Rings famously corrupts the minds of all who wear it, slowly bending them towards evil.
uncertain
who → remember → evil
Just as Frodo Baggins eventually grew to despise the Ring, one of Anthropic’s earliest investors seems to rue the day he first invested in Anthropic (and before it, in OpenAI).
asserted
he → grow → OpenAI
In a statement to the New York Times, Facebook co-founder Dustin Moskovitz wrote: “Cari and I invested in Anthropic to promote safety, not racing.”
asserted
Cari → write → racing
…and 18 more, not listed.