Anthropic reports fewer junior staff as AI agents work more

By Splitfeed EditorialLast updated Oct 18, 2026, 04:50 UTC

Anthropic conducted an internal workforce survey covering 1,334 employees and published the results around October 16 and 17, 2026. The headline figures are striking: AI agents at the company are now logging six times more hours than human staff, the workforce has grown less junior in composition over the past year, and employees are spending more time in meetings than before. The surveyed employees also reported stronger-than-average confidence that AI will displace jobs broadly across the economy, a notable finding given that these are workers at one of the leading AI development companies.

The findings matter because they offer a rare internal data point from inside an AI lab about how the technology is reshaping work at the companies building it, not just at the businesses adopting it downstream. The contraction in junior roles is particularly significant for the wider industry, since entry-level positions have traditionally served as the training ground for the next generation of senior engineers and researchers. Whether the shift in workforce composition is primarily driven by AI agents absorbing tasks previously handled by junior staff, by deliberate hiring strategy, or by some combination, is not settled by the data as reported. The six-times-more-hours figure for AI agents also raises questions about how those hours are being counted and what categories of work they represent.

The anti-AI camp has responded with a range of concerns. Laurie Voss, writing as seldo.com, argued that Anthropic does not hire junior people and has no system for growing new hires into senior ones, and warned that this pattern is already spreading across the tech industry. Others in this camp extended their criticism further, with Elon Green alleging that Anthropic used his book without permission, and Prospect Magazine highlighting a lawsuit by author Ian Rankin. Ed Zitron pointed to what he sees as overinvestment in data center infrastructure, and Maya Media framed the company's situation as a bubble on the verge of collapse. The pro-AI camp has not yet published reactions to this specific story; typically, advocates for AI development would argue that agent-driven productivity gains free human workers for higher-value tasks and that workforce evolution of this kind is a natural and positive consequence of technological progress. The middle ground, represented by Leah McElrath, treated the survey primarily as evidence of the internal pressures AI companies themselves are navigating, stopping short of a stronger judgment.

The argument is likely to sharpen as more AI companies release comparable internal data, or as industry-wide hiring figures for junior technical roles become available over the next year. Any future Anthropic IPO filing would also be expected to shed additional light on workforce structure and the financial role AI agents play in the company's operations.