Defining pattern

This week’s data describes a labor market that isn’t shedding workers so much as standing still. Announced layoffs are at a multi-year low, AI’s share of those cuts is falling, and millions of openings remain posted. But hiring has stalled, more than a quarter of unemployed people have been searching for six months or longer, and the pressure is building in the spaces between jobs: getting a first role, moving into a new occupation, proving real skills, and getting back in after a layoff.

1. Low Layoffs, Long Searches: September’s Jobs Picture

Last week we looked at early signs of a “stuck” labor market. This week’s federal data fills in the picture. The U.S. Bureau of Labor Statistics reported that employers added 29,000 jobs in September, compared with an average of 45,000 a month over the prior year, and revised July and August down by a combined 60,000. Unemployment was 4.2%. August JOLTS data, released September 29, showed about 7.1 million job openings, with hires (5.2 million), quits (3.1 million), and layoffs (1.6 million) all little changed.

  • 1.9 million people have been unemployed for 27 weeks or longer, 27.1% of everyone unemployed.
  • Unemployment for Black workers rose to 7.0%, and 4.5 million people are working part time because they can’t find full-time hours.
  • Financial activities employment is down 129,000 since May 2025, mostly in insurance carriers; BLS does not attribute the decline to AI.

This is a low-hire, low-fire market. People who already have jobs are relatively protected. People who need one face longer searches, even without a layoff wave.

Impact: Career centers should prepare clients for longer search cycles, and time-to-placement may be a more useful signal right now than the unemployment rate.

2. AI’s Share of Announced Layoffs Falls as Hiring Plans Hit a 15-Year September Low

Challenger, Gray & Christmas counted 43,281 announced job cuts in September, down 18% from August and the lowest September total since 2022. AI was cited in 3,961 of those cuts, about 9% of the month and fifth among stated reasons, after accounting for 25% to 40% of monthly cuts from March through July. Year to date, AI has been cited in 120,136 cuts, about 21% of the total. Announced hiring plans of 90,787 were the weakest for any September since 2011.

  • Technology led all industries with 10,799 September cuts, up 77% from August.
  • Nonprofits announced 4,742 cuts, their highest monthly total of 2026, up from 672 a year earlier.
  • Announced cuts are not completed layoffs, and the stated reason is the employer’s own framing.

The “AI layoffs are accelerating” headline doesn’t match this month’s numbers. The weaker signal to watch is hiring, not firing.

Impact: The spike in nonprofit cuts is a warning for the community partners that refer and support job seekers.

3. McKinsey: 11 Million Workers May Need New Occupations by 2035

The McKinsey Global Institute’s base case estimates that automation could reduce labor demand by the equivalent of about 36 million jobs by 2035, while AI-related and broader economic growth could create demand for more than 40 million. The central concern isn’t net job loss. Roughly 11 million workers (a range of 6 million to 16 million) may need to change occupations, with moves across occupational groups running about 3.6 times the historical rate. Only about 14% of those transitions qualify as direct pathways with strong skill overlap, preserved wages, and less than six months of training.

  • More than 70% of projected declining employment is in office and administrative support, retail and sales, and transportation and logistics.
  • Most declining employment is concentrated in the bottom two wage quintiles.
  • Demand for AI fluency in job postings is up elevenfold since 2022, and demand for adaptability is up fivefold.

In this model the bottleneck is mobility, not the number of jobs. That makes it a workforce-system problem: recognizing transferable skills, funding paid training, and building realistic paths from one occupation to the next.

Impact: Training is most useful when it’s built around specific occupation-to-occupation moves rather than generic AI courses.

4. When AI Polishes Every Résumé, Employers Struggle to See Real Skills

A Western Governors University survey of 3,128 U.S. hiring professionals, conducted by Centiment from June 24 to July 7, found that 60% say AI has made candidates’ real skills harder to evaluate. Nearly one in five named confirming that they were interviewing a human rather than AI as a top challenge. Among employers reporting that evaluation difficulty, 54% said AI has reduced entry-level hiring at their organization, compared with 20% of employers who did not report it.

  • The share of employers still figuring out how to assess AI skills doubled, from 16% in 2025 to 32% in 2026.
  • 38% see internships, apprenticeships, or project-based work as the most effective bridge from school to a job.
  • The entry-level finding is an association in a sponsor-funded survey, not proof that AI caused the cuts.

When employers can’t trust the paper, they tend to lean toward experience. That puts candidates without a track record at a disadvantage unless they can show and explain their work live.

Impact: Job seekers should be ready to walk through their process, including how they used AI. Employers can rebuild trust with work samples and structured interviews.

5. New York’s Entry-Level Office Pathways Narrow as AI Skill Requirements Rise

A Partnership for New York City report, using Lightcast job-posting data, found that annual entry-level postings in New York City have fallen since 2022 by 40.6% in design, media and writing; 34.4% in customer and client support; 30.5% in clerical and administrative work; 26.8% in business management and operations; and 23.4% in finance. Entry-level postings that mention AI skills rose 55% over the same period. Separately, Revelio Labs’ September tracker found that at firms adopting AI, senior headcount has grown 32% since late 2022, compared with 6% for junior roles.

  • AI companies leased more than 2.2 million square feet of New York City office space in the first half of 2026, more than double their total for all of 2025.
  • Customer support and clerical work, traditional first jobs for career starters, show some of the steepest declines.
  • Revelio notes its data overrepresents white-collar workers and large cities, and its adopter comparison is not a causal estimate.

AI investment is booming in New York, but it isn’t flowing proportionally into first jobs. Growth is concentrating among people who already have experience.

Impact: Colleges and workforce providers should treat AI literacy as an entry requirement for office careers, not an advanced specialty.

6. Update: New Research Challenges the “Jobless Grads” Story, but Not the Pay Gap

Following up on the Census Bureau study featured in our September 19 edition: a new NBER working paper by Fairlie and Wu of UCLA, using Current Population Survey data, found that unemployment for recent college graduates did not spike in summer 2026 compared with earlier summers, or relative to older graduates and young workers without degrees. The rate for graduates ages 22 to 25 was 7.3%. The Census study, using administrative records, found that graduates in the most AI-exposed majors saw initial employment fall 5 percentage points and initial earnings fall 13%.

  • The two papers may be measuring different things: whether graduates have jobs versus what those jobs pay.
  • The Census researchers found many exposed graduates moved into lower-paying industries rather than becoming unemployed.
  • Fairlie and Wu caution that the classes of 2027 and later could be more affected if workplace AI use keeps growing.

The risk shows up less as joblessness and more as a weaker first job and lower starting pay, which can compound over a whole career.

Impact: Placement counts can miss underemployment. Wage at placement is worth tracking alongside it.

7. AI Use Climbs While Access to Training Falls

PwC’s Global Workforce Hopes and Fears Survey 2026, covering 49,364 workers in 48 countries and territories, found that 64% of workers used AI at work in the past year, up 10 percentage points, and daily generative AI use rose from 14% to 22%. Over the same period, the share saying they have access to the learning and development they need fell from 59% to 51%. Among the 56% of workers PwC calls the “engine room,” fewer than 40% report adequate access to development.

  • The survey was fielded in May and June 2026 across 29 sectors.
  • The training gap is widest for core operational and frontline staff, the “engine room” segment.
  • This is a global survey, so results won’t map exactly onto any single U.S. labor market.

This extends the question we asked on September 12: who pays to become AI-ready? For a growing share of workers, the answer appears to be the worker.

Impact: Employers buying AI tools without expanding development risk widening skills gaps inside their own organizations.

8. California Limits AI-Only Firing and Requires AI Layoff Disclosure

On September 30, California enacted SB 947, the “No Robo Bosses Act,” which bars employers from relying solely on an automated decision system to discipline or fire workers; a human must corroborate the system’s output. It takes effect July 1, 2027. SB 951 adds requirements to California’s WARN notices when a mass layoff, relocation, or termination is caused in whole or substantial part by AI or automation: the notice must identify the affected jobs and locations, the job functions being automated, and the type of technology involved, starting January 1, 2027. Two workplace surveillance bills were signed alongside them. In Connecticut, a provision effective October 1 makes clear that using automated employment decision technology is not a defense against a discrimination claim.

  • A broader version of the No Robo Bosses Act was vetoed in 2025; this narrower version was signed.
  • California’s employment agency will publish summaries of AI-attributed layoff notices.
  • Reported thresholds for SB 951 vary across sources, so the enrolled bill text is the reference to check.

Disclosure only works if employers use it. New York’s existing WARN technology checkbox reportedly had no employer selections as of January, which is a useful caution for how much these notices will reveal.

Impact: Layoff notices that name AI would give rapid response teams earlier and more specific signals about which job functions are being automated.

9. New York’s FutureWorks Commission Sets a December Deadline

On September 29, New York State convened its FutureWorks Commission on AI and the workforce, following statewide listening sessions with workers, unions, and experts. The commission is scheduled to deliver specific recommendations by the end of December on data and forecasting, job training and placement, job quality, social safety nets, economic growth, and support for small businesses.

  • The state named administrative, clerical, and customer-service roles as examples of jobs whose tasks could change.
  • No Finger Lakes-specific proposals have been announced, and the final recommendations aren’t public yet.
  • This continues the listening-session process we covered on August 22.

This is policy development, not a new protection or funding stream yet. But New York’s response to AI displacement is being written right now.

Impact: Local workforce boards and providers that want their evidence reflected have a short window before December.

10. New York’s Unemployment System Goes Offline for Up to 12 Days in November

The New York State Department of Labor will take all unemployment insurance services offline for nine to 12 days starting November 2 at 5 p.m. to replace its 50-year-old system, the final step in a decade-long, $180 million project. The new system adds smartphone access and real-time claim tracking, and backdated filings will be allowed once it’s live.

  • Claimants should certify for the week ending November 1 before the 5 p.m. cutoff on November 2.
  • Phone services are expected to be offline as well, and claimants must still complete and record work search activities during the outage.
  • Career Centers are expected to remain available virtually by appointment.

The upgrade is long overdue, but the gap lands on people already navigating a slow hiring market, and hardest on those with limited digital access.

Impact: American Job Centers should prepare for extra traffic and questions in November. Spanish-language support plans for the outage could not be verified.

11. A Fed Governor’s Warning: Interest Rates Can’t Fix an AI Skills Mismatch

In a September 28 speech, Federal Reserve Governor Lisa Cook said evidence that AI is structurally reshaping the labor market is still limited, pointing to signs of reduced demand in software coding and simultaneous translation and to recent graduates struggling to land first jobs. She described a scenario in which AI raises unemployment because workers’ skills no longer match available jobs. In that case, she said, the Fed would have limited tools, and cutting rates could fuel inflation. She noted that fiscal policy could help but is outside the Fed’s purview.

  • She cited Fed survey data showing nearly half of small employer firms use AI, and 71% of those report higher productivity.
  • Twelve-month inflation reached 3.8% through August, above the 3.0% wage growth in the September jobs report (a comparison of two published figures, not an official real-wage calculation)
  • These are a policymaker’s views, not a forecast of Fed action.

In effect, a central banker is saying that if AI creates a skills mismatch, the main response will have to come from training and the workforce system.

Impact: This is a citable argument for workforce funding at a time when federal appropriations remain unsettled.

12. Robots Can Do Much of Physical Work, but Rarely More Cheaply Than People

Anthropic research estimates that robots can technically perform 74% of physical tasks, representing 34% of U.S. working hours, in at least some settings. But robots are currently cost-competitive with human labor for only 0.3% of job tasks, and at historical rates of robot price decline it would take about 40 years to reach 10%. Driving and warehouse work are among the most exposed; nursing and general repair are among the least.

  • Highly robot-exposed workers are 16 percentage points more likely to be Hispanic or Latino and 55 points less likely to hold a bachelor’s degree than unexposed workers.
  • Drivers, warehouse workers, packers, and stockers sit at the top of the exposure list.
  • This is company-authored research based on task and cost models, not observed displacement.

“Can perform” is not the same as “will replace.” The gap between technical capability and economically viable deployment is enormous, and it complicates claims of rapid physical automation.

Impact: Logistics and transportation training should prepare workers for job redesign in controlled settings like warehouses, without assuming mass replacement is imminent.

13. Training Money Moves Toward the Trades, Where the Pipeline Falls Short

The U.S. Department of Labor awarded $43 million in Industry-Driven Skills Training Fund grants to seven states for employer-driven training in shipbuilding, advanced manufacturing, aerospace, nuclear energy, cybersecurity, and IT. Across both rounds, the program has now directed more than $129 million to 21 states. Separately, the Alliance for America’s Skilled Trades released its first report covering 124 skilled-trades occupations, projecting 1.7 million openings a year through 2035 while current training pathways produce 55 workers for every 100 needed.

  • More than $30 million of the fund’s total has gone to shipbuilding-related training.
  • Only about half of people who enroll in skilled-trades training programs finish, according to a coalition partner’s announcement.
  • The Alliance’s figures come from an industry coalition and its research partners, not an official forecast.

As some routine office pathways narrow, growth is showing up in hands-on technical work that doesn’t require a four-year degree. The bottleneck there isn’t demand; it’s completion.

Impact: Pre-apprenticeship, completion supports, and paid training are capacity issues, not extras, for programs feeding the trades.

Bottom line

The loudest story about AI and work is still about replacement. This week’s evidence points somewhere quieter: movement. McKinsey’s 11 million occupational transitions, employers struggling to verify real skills, New York’s narrowing entry-level pathways, and a Fed governor saying interest rates can’t fix a skills mismatch all land in the same place, the space between one job and the next.

That space is where workforce development lives. Whether it’s a career center bracing for a November benefits outage, a training program built around a specific occupational move, or a hiring manager replacing résumé screens with work samples, the work that matters most right now is helping people move, and helping them prove what they can do once they get there.