AI & Work
Anthropic has sharply reduced the share of junior people among its technical hires, according to a report the AI developer published on Friday, Oct. 16. Junior hires fell from 27% to 9% of the company’s technical hiring since early 2025, while the junior share of employees in its technical departments slipped from 21% to 14% over the same stretch.
The report, titled “How employment is changing at Anthropic” and written by Nathan Wilmers, Regina Catipon, Maxim Massenkoff and Peter McCrory, uses the company’s own headcount, hiring and usage records to describe how work has changed as its staff adopted AI agents. The authors frame Anthropic as a possible leading indicator of how AI could reshape work organization, hiring and labor demand, while cautioning that many of its employment practices cannot be replicated across the economy.
What the numbers measure
The figures describe the mix of hiring, not job losses. The report says technical hiring “has continued growing in absolute terms,” but that hiring in sales, finance and recruiting grew faster as business functions expanded with the company, which the report says has tripled in size in the last year. As a result, technical departments’ share of total employment fell from 61% to 56% over the past year. The report does not mention layoffs and says that, as jobs evolve, the company has prioritized internal redeployments.
The report does not give a formal definition of “junior,” referring to junior-level and lower-band technical positions without specifying levels, and it does not state an end date for the period it covers. It adds that job postings for junior roles have fallen particularly sharply since December 2025. Descriptively, the authors say, junior employment slowed most in jobs that use the company’s AI for the most complex tasks.
Agents logging more hours than people
The shift in hiring coincided with a rapid rise in AI agent activity inside the company. According to the report, combined autonomous and interactive agent work hours passed human work hours for the first time in March 2026, and by August interactive agent hours outnumbered employees’ hours six to one. That comparison has limits: human hours are estimated from headcount at an assumed 40 hours a week, and an hour of agent activity is not the same as an hour of human work.
The report also documents output gains. Code commits per engineer were running at five times their level of six months earlier. Coding sessions by nontechnical staff rose from two a month in August 2025 to 90 in May 2026. The authors tracked 192 engineers as they adopted coding agents and compared them with 74 who did not, but caution that adoption was not random and that adopters were already on a steeper upward trajectory.
A different job, and different hiring criteria
Much of the new work is supervision. Reading all the output of their AI agents would take nontechnical employees about half their work week, the report estimates, and so-called meta-work, such as building monitoring tools and overseeing agent pipelines, rose from 10% of the average employee’s interactive sessions in fall 2025 to 22% in spring 2026. One in five surveyed employees described their role shifting toward managing AI agents rather than doing the work directly.
Hiring criteria have moved too. Hiring managers noted candidates’ track record with AI tools in about 5% of technical-role assessments in 2024; by 2026 that rate had tripled for technical roles and quadrupled for nontechnical roles. The biggest increases were in comments about applicants’ autonomy and judgment, while discussion of domain expertise shrank. Around a third of machine-learning managers and a fifth of software-engineering managers said they were hiring less because of AI.
What it means for early-career workers
The report describes one fast-growing AI company with unusual access to its own tools, and its authors say lessons for other employers should be drawn carefully. It is still a concrete, company-level record of the pattern that early-career technology workers have worried about: fewer junior openings, more weight on judgment and on the ability to direct and check AI output, and continued demand for experienced staff. Whether other employers follow the same path will show up in their own hiring data, not in a single company’s report.
Sources
- Anthropic, “How employment is changing at Anthropic,” October 16, 2026. https://www.anthropic.com/research/employment-at-anthropic
Featured photo: Colorful software code on a computer monitor, photographed Feb. 14, 2017. Representative image; not Anthropic's offices, systems or employees. Photo: Markus Spiske, via Wikimedia Commons (CC0). Cropped from the original.
.jpg){kind=link}