The Youngest Workers in AI’s Path Have Lost Ground for Three Straight Years
Sign Up For Daily Job Market News
A new data update this week puts a number on something many young job seekers have felt for a while: the jobs AI touches most keep shrinking for people just starting out. ADP Research and the Stanford Digital Economy Lab jointly manage a tracker called the Canaries Dashboard, built to test whether AI-exposed jobs are losing workers faster than jobs AI can’t easily touch. The August update draws a clear line.
Across all workers, employment in the most AI-exposed occupations fell 0.6% from a year earlier, while employment in the least-exposed occupations grew 0.2%. This gap widens sharply once you isolate the youngest workers in the data. Employment for workers ages 22 to 25 fell 3.1% overall. Within that group, employment in high-exposure jobs fell 4.4%, marking the 35th straight month of year-over-year contraction in a streak that started in October 2023, 12 months after ChatGPT’s public release.
Young workers in the least-exposed jobs lost ground too, down 2.0%, but at less than half the pace of their peers in AI-exposed roles. Workers ages 26 to 30 show a similar split: down 3.1% in high-exposure jobs against a gain of 0.1% in the least-exposed ones. The data comes from ADP payroll records covering millions of workers across more than 730 occupations at tens of thousands of private U.S. employers, updated every month.
Thirty-five months in a row is long enough to call it a pattern, not a blip. Employment for the youngest workers in the jobs AI touches most has shrunk every single month since October 2023, and the gap keeps widening instead of closing. I’d also point to the other number buried in this report. Young workers in the least-exposed jobs lost ground too, down 2.0%. AI explains part of the problem, but entry-level hiring is weak across the board, and AI-exposed roles are taking an additional hit on top of that.
The 26-to-30 group worries me more than the headline number does. Those workers aren’t brand new to the labor force anymore, yet they’re still down 3.1% in AI-exposed jobs while their peers in other roles are flat. If companies use AI to skip the first rung of the ladder, the second rung starts disappearing a few years later, exactly the concern I raised after a similar Stanford finding earlier this week showing senior hiring rising while junior hiring falls at AI-adopting companies.
For employers, the math may look good today, but you still need a bench. The person you don’t hire at 23 is the manager you won’t have at 30. For young job seekers, the data points toward roles built on hands-on work, relationship-building, or the kind of judgment that’s hard to give a tool. If your team’s entry-level pipeline has quietly dried up, talk to us about hiring and we’ll help you rebuild it with intention instead of by accident.
Staffing your team doesn’t have to be hard.
Reach out and see how we can help.
The Fed Just Measured How Frozen Hiring Is
Everyone keeps calling this a “low-hire, low-fire” job market, but the Richmond Fed put an actual number on how unusual it is. Economists Claudia Macaluso and John O’Trakoun used federal JOLTS data going back to December 2000 to measure where today’s market falls against every month in the last 26 years.
Looked at individually, each number seems low but not alarming. The July hiring rate was 3.2%, and only 7% of months since December 2000 came in lower. The July layoff rate was 1.0%, and only 4% of months were lower. COVID skews the layoff history badly here: the rate hit 8.5% in March 2020 and 6.9% in April 2020, numbers that make today’s reading look almost calm by comparison.
The bigger finding shows up when you look at both sides together. The authors grouped 12 measures into a hiring channel, 8 measures including job-finding odds, new hires, openings, quits, and one-month changes in payrolls, temp help, and the employment-to-population ratio, and a separation channel, 4 measures covering job losers, initial jobless claims, the layoff rate, and the odds an employed worker becomes unemployed. Each measure gets ranked against its own history since 2000.
The latest reading sits at the 43rd percentile for hiring and the 90th percentile for separations. Getting hired is a little harder than usual. Losing your job is close to the least likely it has ever been. Since 2023, readings have drifted into the weak-hiring, low-separations corner of the map, a spot above every other month measured since 2000. October 2007, two months before the Great Recession, offers a useful contrast: hiring then sat near the 49th percentile, close to today’s reading, but separations were near the 38th percentile and already climbing. Separations today are flat and lower than they were even in mid-2016, well into an expansion. The hiring channel has been below the 50th percentile for more than 2 years, though August did show a small improvement, rising from the 36th percentile in July to the 43rd.
People keep calling this market frozen, and the Richmond Fed just put a picture to the word. Companies are keeping the people they have and adding few new ones, and both are happening at once, putting this market further from normal than any other point in the last 26 years. If you have a job, your odds of losing it are about as low as they’ve ever been. If you’re trying to get in, or trying to move, the door is much heavier. The hiring side has been below average for more than 2 years, and that gap explains why so many job seekers feel stuck even though big layoff headlines stay rare.
I’d put the October 2007 comparison in front of any executive worried about a recession. Hiring back then looked a lot like it does now, but separations were already climbing. Today they’re lower than they were in mid-2016, when the economy was clearly expanding, so this data doesn’t resemble the setup for a downturn. For employers, low turnover feels great right up until you need to hire. The people you want are staying where they are, so pulling them away takes a better offer and a faster process than this market has demanded in years. August’s move from the 36th to the 43rd percentile is one month of data, so I’m not calling it a thaw yet, but it’s worth watching the next release closely.
Fake Recruiters Are Getting Bolder, and So Is the Man Hunting Them
A story in The Wall Street Journal this week puts a face on a problem most job seekers have already encountered somewhere: fake job ads and fake recruiters. Jay Jones, a 39-year-old unemployed copywriter from the Chicago area who goes by “The Profiler,” spends hours every day finding fake job ads and fake recruiter profiles, mostly on LinkedIn, and reporting them until they come down. He’s tracked nearly 60,000 fake jobs and flagged about 7,000 scam profiles so far.
Jones got into this after his own layoff from an advertising job, when a fake resume writer asked him to pay early in his search and he started digging instead of paying. Scammers now use AI to tailor fake roles to a specific person’s profile, with a common pitch that goes something like “Your resume isn’t good enough, but pay up and we’ll get you an offer.” Some scammers ask for a driver’s license or Social Security number outright and sell the data.
One tactic stands out: scammers hijack forgotten LinkedIn accounts that still hold a verified badge and relist them as recruiters at real companies. In one case from the article, Kelley Barcus, CEO of a one-person Colorado recruiting firm called Argenta Talent Acquisition, found hijacked profiles listed as her own employees, and LinkedIn gave those profiles 14 days to update or appeal before removing the false link to her firm. In another, roughly 30 fake roles reportedly appeared on Hyundai Motor’s verified LinkedIn page, linking out to a non-Hyundai site, and fake Capital One “Administrative Assistant” ads led to a similarly unaffiliated site.
This is landing on a workforce that’s already stretched thin. The share of unemployed people searching for more than 6 months was 27% in August, near its highest level in more than a decade outside the pandemic. LinkedIn says its AI systems catch nearly all scams before members ever see them, and that it stopped close to 90 million fake accounts before they went live in the second half of 2025. The company also partnered with the FBI this summer on job seeker education and, later this month, plans to roll out a feature letting verified employers instantly remove false affiliations from their page.
Scammers go where the desperate people are. With 27% of unemployed people searching for more than 6 months, a message promising “I can get you an offer” hits a lot harder than it would in a hot market. The rule I’d give every job seeker is simple: a legitimate recruiter never asks you for money. Employers pay recruiters, not the other way around. If someone wants a fee to fix your resume or get you an interview, or asks for your Social Security number before you’ve even spoken to a real person, walk away.
The Barcus example should get the attention of every staffing firm owner. Scammers borrowed a real recruiting firm’s name to look credible, and the platform’s process still gave them two weeks to keep working before the link came down. Every fake recruiter makes candidates more suspicious of the real ones, and that cost lands on all of us in this industry. For employers, the Hyundai and Capital One examples show your own verified company page can display ads you never posted. Someone on your team should be checking what’s showing up under your name, and your careers site should be easy enough to find that a candidate can verify a role in 30 seconds flat. If you’re a job seeker worried about spotting the difference, see what’s open right now through a source you can verify directly.
