The woman in this opening scene does not exist; she is a composite, built for this article from the kinds of cases the Federal Reserve’s contacts describe. Call her a graduate in accounting. She finished her exams in the spring, signed an offer from a large firm in Dallas and was told to expect a start date in January. This week an email arrives from a recruiting coordinator. The firm is “revisiting the timing” of new contracts. Nobody has been fired, because nobody was ever employed. A seat that was going to exist simply does not.

Her case is invented, but the pattern is on the record. On 14 October the Federal Reserve published its Beige Book, a summary of what business contacts told the twelve regional Reserve Banks in the weeks up to 5 October. It is anecdote, not measurement, and the Fed says so: the document summarises comments from contacts outside the system and is not a statement of the views of Fed officials. Still, one sentence in it is new. “Five Districts reported that some firms had either reduced staffing levels (through attrition), delayed hiring, or anticipated less job creation after increasing productivity by investing in AI.”

Three editions, one changing sentence

The Beige Book appears several times a year, so the wording can be followed like a thermometer. The July edition, published on 15 July, said that “a few Districts noted that firms had increased their usage of AI, either in the hiring and screening of potential employees or to boost worker productivity.” One district added that employers “held head counts steady and invested further in AI.” In that report employment rose on balance, with five districts showing gains.

The August edition, released on 2 September, was vaguer and more divided: “Districts also reported both positive and negative effects of artificial intelligence on labor demand.” It did not say where the negative effects fell. October says it. The subject of the sentence is no longer a tool that firms use but a headcount that firms do not fill.

The overall labour market, in the same October report, was “little changed”. Three districts described modest employment gains, five slight gains and four no significant change. Shortages persisted for specialised, technical and skilled-trade workers. Whatever AI is doing to hiring, it is doing it inside a labour market that is not falling apart, and the report’s own price section is dominated by diesel, tariffs and insurance rather than software.

What the district reports describe

The detail sits in the individual district pages. In Chicago, a business-services contact said productivity gains from AI adoption had let the company reduce employment through attrition. In Cleveland, two banking contacts said automation and AI had reduced staffing needs by raising the productivity of certain roles. In New York, some larger organisations were beginning to cut selected jobs or hold off on hiring as they integrated AI, and demand for marketing roles softened, though there were no reports of large-scale layoffs.

Dallas offered two of the sharpest accounts. An accounting contact noted that some large firms are delaying new employment contracts because they may not need as many entry-level employees as first expected, partly because of efficiency gains from AI. A staffing firm described the other side of the ledger: it had hired new recruiters to manage its AI expansion, after AI agents conducted more than 180 hours of interviews in two days and people were needed to review the results and decide.

Minneapolis supplied the formulation that fits the whole report best. IT staffing contacts said artificial intelligence “was not causing layoffs but was dampening the creation of new jobs.” Atlanta, meanwhile, reported that AI implementation had broadened as more contacts deployed tools and automation, and that, as in earlier reports, most firms did not expect significant workforce reductions in the near term. San Francisco heard of consultants leaving large, legacy firms to start AI-enabled businesses of their own. Cleveland added a cost note: banking and professional-services contacts reported rising costs for software, AI tools and computer hardware.

The payroll data point the same way

A Beige Book cannot say how large any of this is. Another source can say something about who is affected. The Stanford Digital Economy Lab published a revised version of “Canaries in the Coal Mine?” on 12 August. Erik Brynjolfsson, Bharat Chandar and Ruyu Chen use payroll data from ADP that cover millions of American workers through June 2026. They find no evidence of widespread, economy-wide job displacement. They do find that employment of workers aged 22 to 25 in AI-exposed occupations now stands 19% below where it would be had it kept pace with less-exposed peers, while experienced workers show no comparable gap.

The mechanism the authors report matches what the Fed’s contacts describe: the gap operates mainly through reduced hiring of young workers rather than increased separations. It is concentrated in occupations where AI usage mainly substitutes for human tasks; where usage mainly complements workers, employment is flat or rising. The authors first documented the divergence in August 2025 and say it has widened steadily since. They are careful about what it proves. They call the results early, descriptive indicators, not causal estimates, and note that the patterns weaken when education is controlled for and that some divergent trends predate generative AI.

Put the two sources side by side and the tension in the story becomes visible. Firms tell the Fed that AI is not causing layoffs, and in the narrow sense that appears to be true. Both the Fed’s contacts and the payroll data suggest the adjustment is happening at the door instead: fewer offers, later start dates, vacancies left unfilled when someone retires. A person who is never hired does not show up in a layoff announcement, a severance figure or a headline.

What the evidence cannot carry

The limits are real. The Beige Book covers five districts out of twelve on this point, and each gave a few cases, not a count. It names no sector totals and puts no number on lost jobs. Its contacts attribute the changes to AI, and attribution by an employer is a claim, not a finding. The same report cites interest rates, tariffs, fuel prices and uncertainty as reasons firms hold back, and Cleveland’s real-estate and construction contacts trimmed staffing for that reason alone. Dallas, meanwhile, reported that AI agents created work for human reviewers.

What the report does establish is a direction of travel in an official, regularly repeated survey. In July the AI sentence was about adoption. In October it is about people not hired. Whether five districts becomes more is a question for the next edition.

Back in Dallas, the composite graduate answers the email within the hour, politely, and then opens a spreadsheet of other firms. Her draft cover letter now has a new line about the tools she used in her last internship. She is not out of work. She is where the Fed’s contacts say the pressure lands first: at the front of a line that has quietly got shorter.

References

  • Brynjolfsson, E., Chandar, B., & Chen, R. (2026, revised 12 August). Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence. Stanford Digital Economy Lab. digitaleconomy.stanford.edu
  • Federal Reserve System (2026). Beige Book, July 2026: National Summary. federalreserve.gov
  • Federal Reserve System (2026). Beige Book, August 2026: National Summary. federalreserve.gov
  • Federal Reserve System (2026). Beige Book, October 2026: National Summary. federalreserve.gov
  • Federal Reserve Bank of Atlanta (2026). Beige Book, October 2026: Sixth District. federalreserve.gov
  • Federal Reserve Bank of Chicago (2026). Beige Book, October 2026: Seventh District. federalreserve.gov
  • Federal Reserve Bank of Cleveland (2026). Beige Book, October 2026: Fourth District. federalreserve.gov
  • Federal Reserve Bank of Dallas (2026). Beige Book, October 2026: Eleventh District. federalreserve.gov
  • Federal Reserve Bank of Minneapolis (2026). Beige Book, October 2026: Ninth District. federalreserve.gov
  • Federal Reserve Bank of New York (2026). Beige Book, October 2026: Second District. federalreserve.gov
  • Federal Reserve Bank of San Francisco (2026). Beige Book, October 2026: Twelfth District. federalreserve.gov