Economists at Stanford's Digital Economy Lab, led by Erik Brynjolfsson, have been running an ongoing study nicknamed "Canaries in the Coal Mine," tracking what happens to young workers in roles most exposed to generative AI. Their August 2026 update, built on payroll data covering tens of millions of US workers, found that employment for 22-25 year-olds in the most AI-exposed occupations now sits roughly 19% below where it would be expected to track, based on how less-exposed peers have fared. A year earlier, that gap was closer to 13-15%, meaning the divergence has been widening rather than stabilizing.

It's about hiring, not layoffs

The mechanism the researchers describe is specific: the decline is driven mainly by employers opening fewer entry-level positions in AI-exposed roles, not by companies laying off the young workers already in those jobs. Experienced workers doing the same kind of work show no comparable drop in employment. In other words, the people already inside these roles aren't losing their jobs at unusual rates — it's the door into those roles that's narrowing for new graduates.

The study's authors are careful to call their findings "descriptive, not causal," and a separate analysis from the New York Fed has pushed back on how cleanly the AI link can be established versus other explanations, like a broader slowdown in entry-level hiring across the economy. It's a real and measurable pattern, but "AI-exposed roles are hiring less" is a more accurate description than "AI is directly replacing entry-level workers."

A similar pattern shows up elsewhere

A separate compensation report from Ravio, covering the European tech sector, found junior tech hiring rates fell by roughly 73% year-over-year even as the volume of new AI and machine-learning hires grew by around 88% in the same market. Different methodology, different geography, but the same shape: hiring is concentrating around AI-specific or more senior roles, while general entry-level hiring in exposed functions is shrinking.

What this means if you're early in your career

• Generic entry-level roles in highly AI-exposed functions (first-pass content writing, basic coding support, routine data entry and analysis) are the hardest to break into right now — competition for the remaining openings is intensifying.
• Specialization and demonstrable work matter more than they used to. A portfolio, a relevant internship, or a specific technical credential helps you stand out in a shrinking pool of entry-level openings.
• Roles that pair domain expertise with AI tools — rather than roles AI tools can substitute for outright — are where hiring is holding up or growing.

None of this means entry-level hiring has collapsed everywhere; it's concentrated in roles where AI tools most directly substitute for junior-level tasks. But if you're graduating into one of those fields, it's worth treating the data as a signal to differentiate early rather than assume the market will look the way it did for the cohort before you.