Anthropic Studied Its Own Workforce, and One Number Should Scare Every New Grad

Interactive AI agents now log six hours for every human hour. The humans got more meetings.

THE ENTRY-LEVEL CUBICLE, RECENTLY VACATED, SAN FRANCISCO (The Daily Whatever Press)—Last week, Anthropic did something tech companies almost never do. It pointed its own research team at itself and published what it found.

What it found is that the robots are eating the interns. Probably not literally.

The report, “How employment is changing at Anthropic,” came out Oct. 16 from researchers at the Anthropic Institute. Since early 2025, the share of junior people in the company’s technical departments fell from 21% to 14%. The junior share of technical hires dropped from 27% to 9%. That’s a two-thirds drop in the junior share of new technical hires, at the company whose CEO has spent more than a year warning the rest of us that this exact thing was coming.

Eek.

The numbers in the report read like an episode of Office Space written by the printer. AI agents passed humans in work hours at Anthropic in March 2026, and by August interactive agent hours outnumbered employees’ six to one. Code commits per engineer are up about fivefold from six months earlier. About one-third of machine-learning managers and one-fifth of software managers said in an April survey that they’re hiring less because of AI.

The quotes are better than the charts. One person quoted in the report said a jump in what the newest models could do “basically overnight changed what we were looking for.” An executive assistant said, “I think we’ve already moved past the point of Claude augmenting away the really entry-level admin type role.” An engineer described working with “an army of Claudes, with every one asking for my attention.”

And my favorite: about 80% of the time in interactive Claude sessions is spent with the AI waiting on a human. The machine is sitting there tapping its foot while we finish reading the email. Meanwhile median meeting time at the company is up 20% year over year, so the humans who survived got more meetings.

If the Bobs from Office Space showed up at Anthropic today and asked a junior engineer what exactly they’d say they do here, the honest answer would be, “Not much, since March.”

One unnamed executive in the report said, “a workforce that is anxiously worried is not a workforce doing its best work.” Correct. The staff expect AI will be able to do 75% of their work by 2030. Other Americans, surveyed for comparison, guessed 28%. The people closest to the machine are the most sure it’s coming for the job.

The CEO saw this coming

In May 2025, Dario Amodei told Axios that the shock to white-collar work is “going to happen in a small amount of time—as little as a couple of years or less.” Axios summarized his warning as AI possibly wiping out half of all entry-level white-collar jobs. He also said, “Most of them are unaware that this is about to happen.”

Not at Anthropic, they weren’t. They had the dashboard.

To the report’s credit, it says out loud that none of the company’s own responses “directly address the slowdown in junior hires,” and it calls for government retraining and income support. That’s an honest sentence. It’s also a company telling the government to clean up after a trend the company is measurably accelerating in its own building.

Awkwaaaard.

Then came the fund

Two days later, on Oct. 18, Anthropic announced the Anthropic Public Benefit Fund. It starts with 5% of the company’s shares, subject to stockholder approval, and the board can add 2% more for every 50% rise in the company’s value above a baseline it hasn’t disclosed. The fund is an LLC, not a charity, and its equity can’t be used for lobbying or political giving.

At Anthropic’s last private valuation of $965 billion, 5% is roughly $48 billion on paper, by my math. That is a serious pile of money, and tech writer Kevin Roose said the broader pledge could be “the largest commitment of philanthropic capital ever made.”

What will it do? Anthropic says it may make “large donations to traditional nonprofits,” make “direct distributions to specific populations,” and that “given the pace of AI progress, bolder approaches may be needed.” Who runs it, how it’s governed, and what the milestones are will be announced “in the coming months.” Nothing has been spent.

The critics lined up fast. Elon Musk said, “I recommend giving the shares directly to the public,” and then promised every American kid a SpaceX share, starting next year, details pending. David Sacks wrote, “You’ve put 20% of your equity in an LLC. Whether it serves the public good remains to be seen.” Gary Marcus said, “Anthropic is playing games here.” Palmer Luckey said it “reeks of USAID 2.0,” which is a strange insult from a guy whose company sells weapons to the government.

The part I mean

I spent 25 years in tech. The whole industry ran on juniors who got to be bad at things on somebody else’s payroll until they were good at them. That’s where senior engineers come from. That’s where program managers come from. You can’t hire a 10-year veteran in 2036 if nobody hired the first-year in 2026.

Anthropic deserves real credit for publishing its own numbers. Most companies would have buried them in a drawer.

But a 27%-to-9% collapse in junior technical hiring is happening right now, to people with names and student loans, and a fund with no governance, no recipients, and no milestones is a promise about later.

Dario Amodei warned us the entry-level jobs would go. His own company just proved him right, and the kids who would have gotten those jobs need more than “the coming months.”

---

Thanks for stopping by!

We’re Dana DuBois and Lawrence Winnerman, lifelong BFFs and co-hosts of The Daily Whatever Show — our independent home for smart, irreverent conversations about culture, politics, media, creativity, relationships, aging, and whatever else is shaping modern life this week.

We’re live every Friday at 10am ET with thoughtful analysis, sharp interviews, cultural commentary, humor, and the kind of conversations that don’t pretend everything is fine when it clearly isn’t.

If you’re enjoying what we’re building, please consider supporting our work with a paid subscription. Your support helps us keep the show independent, book fascinating guests, and continue creating smart conversations with your morning coffee.

We love you — mean it!