Pedagogue Systems
← The Brief

The hiring slowdown has a profile.

Stanford's revised canaries study and a Richmond Fed brief find the decline concentrated in AI-exposed work, healthcare staffing logs four deals in a week, and three regulatory clocks move.

By Pedagogue Systems · August 16, 2026

Two studies found the same hiring pattern in different data.

On August 12, Stanford's Digital Economy Lab released a revised version of Canaries in the Coal Mine, its study of AI and early-career employment. The authors, Erik Brynjolfsson, Bharat Chandar, and Ruyu Chen, now use ADP payroll data through June 2026. The headline finding concerns workers aged 22 to 25 in the most AI-exposed occupations. Their employment is about 19 percent below where it would be had it kept pace with less-exposed peers. In levels, employment for that age group in the two most exposed quintiles fell about 11 percent between November 2022 and June 2026. The three least exposed quintiles grew about 10 percent.

The authors find no evidence of widespread displacement across the economy. They call the evidence descriptive rather than causal. The gaps shrink when controlling for education, and they run larger in the ADP sample than in national benchmarks. The divergence operates mainly through reduced hiring, not separations. It has widened steadily since it was first documented in August 2025.

The Richmond Fed's Economic Brief 26-26, by Katarina Borovickova and Claudia Macaluso, found a related pattern the same week. It splits the labor pool by attachment to work. Primary workers, about 55 percent of the pool, are almost always employed. Their job-finding rate fell 13 percentage points from its November 2022 peak to a September 2025 trough. Secondary workers, who cycle in and out of work, lost only 2 points. The declines are largest in AI-exposed occupations. The authors note this is not the usual recession pattern, in which the secondary segment absorbs most of the adjustment.

Pedagogue Systems' view. Both studies land on the hiring margin, not the firing margin, and both hedge their causal claims. Together they describe weaker hiring access for young workers in highly exposed occupations and for unemployed workers with the strongest prior attachment to work. For operators, that is a demand-mix signal, best tracked by segment.

An AI safety team hedged against its own company's screening.

Per Bloomberg reporting on August 10, Google DeepMind's AGI Safety and Alignment Team encourages job applicants to fill out a separate form that travels alongside the standard application. A document Bloomberg viewed warns candidates that the internal application system carries a "non-trivial probability" that a CV will be screened out incorrectly or take too long to reach the team. The form, it says, ensures a person sees the application. A DeepMind spokesperson denied that the systems filter applicants incorrectly, saying instead that the form routes resumes directly to the team. Google separately sells AI features that help businesses evaluate resumes and forecast hiring needs.

Pedagogue Systems' view. The document and the denial describe the same mechanism differently: one as a hedge against error, the other as a routing convenience. Both agree the form puts a resume in front of a person.

Healthcare staffing logged four acquisition announcements in under three weeks.

On August 11, CHG Healthcare acquired KREWE Anesthesia, a CRNA staffing firm that self-reports about 84 percent annual clinician retention. Kyndryl announced a pending agreement on August 10 to acquire Healthcare IT Leaders, a healthcare IT consulting and staffing firm. Philips International closed its purchase of The Nash Group's healthcare consulting, technology, and recruiting businesses on August 6, picking up scheduling and acuity tools; the buyer is the privately held New York real estate investment and management firm, not the health technology company. And on July 24, Care Career announced what it describes as its seventh acquisition in 24 months: MAS Medical Staffing. The deal adds travel, allied, and per diem staffing plus the Maestra scheduling platform, on which the company says 68 percent of providers self-schedule. No transaction value was announced for any of the four.

The demand backdrop provides context for the acquisitions. SIA's July Pulse survey found median June revenue growth positive across all four healthcare staffing segments it tracks. AMN Healthcare's August 12 report cites the Association of American Medical Colleges. The AAMC projects a physician shortage of up to 86,000 by 2036. AMN's own data show 2025 urban RN fill rates at nearly three times rural rates.

Pedagogue Systems' view. Two of the four deals picked up scheduling software along with the book of business. The segment is returning to modest growth against a projected decade of shortage. The scheduling layer, where clinician allocation happens, is now part of what acquirers are buying.

Three regulatory clocks moved.

On August 11, the Colorado Department of Law filed proposed rules with the Secretary of State. The rules implement the Automated Decision-Making Technology Act (SB 26-189) and the Chatbot Safety Act. Written comments run through October 26. Comments submitted by September 4 will be considered for the revisions presented at that hearing. Both laws take effect January 1, 2027, and the rules would take effect the same day. We covered the statute when it was signed in May.

The same day, the EEOC held its public hearing on the proposed rescission of the EEO-1 through EEO-6 reports. We covered the proposal when it was published. Twenty-two witnesses testified, and a majority urged retention or modernization. Witnesses on both sides noted the same data can still be sought through investigations, subpoenas, and litigation. The commission took no formal action. Written comments close August 24.

In California, SB 1032, the staffing agency registration bill, is dead for the year. The Assembly Appropriations Committee held it on August 13. The bill had cleared four policy committees and passed the Senate floor. The American Staffing Association reported the outcome. It says it campaigned against the bill alongside the California Staffing Professionals for the past year.

Pedagogue Systems' view. The Colorado dates are the operative facts: September 4 for comments to shape the pre-hearing revisions, October 26 for the comment period and the hearing itself. On the EEO-1, rescission would end the scheduled report. Case-by-case demands, through investigations, subpoenas, and litigation, stay untouched; witnesses on both sides made that point. And because the Appropriations hold sidestepped the merits, SB 1032's requirement list could serve as a template for a future bill, in California or elsewhere.

The staffing hours reading held.

Last week's Brief asked whether the 10 percent staffing hours figure would hold. The next SIA and Bullhorn Staffing Indicator answered: hours were up 9 percent year over year for the week ended August 1, indexed at 102, another 2026 high. Industrial led at plus 13, commercial ran plus 10, professional plus 8, IT plus 6, and office and clerical fell 6. Recent weeks are preliminary and subject to revision.

Pedagogue Systems' view. One point off the prior reading, with revisions pending, leaves the reading intact. The segment mix matters more: industrial-led, with office and clerical negative.

What we are watching.

The ASA's AI in Staffing Week runs August 17 through 21. Workday reports fiscal Q2 results on August 27. The BLS preliminary benchmark revision lands August 28. Star Equity's go-shop period for its Harte Hanks acquisition expires September 13. The deal is $38.4 million in aggregate equity value, the release's own label.

About Pedagogue Systems. Pedagogue Systems builds Cassion, a governed data foundation for staffing operations. It serves shift-based, credential-heavy operators in healthcare, industrial, and aviation staffing. It helps staffing operators keep operational records attributable and auditable before they automate decisions.

Sources.

This Brief was produced with AI assistance and adversarial review, then edited by a human before publishing. AI and people both make mistakes, so please verify anything critical independently. How we produce the Brief.

← Back to the Brief