EVIDENCE FILE
Is AI Really Taking Our Jobs - or Creating More Than It Destroys? - Evidence File
The claim register for this article, with each claim's label and source. It records every figure the article asserts and every judgement it makes in its own voice. Generated from the article's own register on 10 November 2026, so this file and the article cannot disagree about the count.
The register at a glance
| Measure | Value |
|---|---|
| Substantive sources retrieved and read | 17 |
| Sources carrying a resolvable URL | 17 |
| Claims registered | 104 |
| Labelled VERIFIED | 83 |
| Labelled PRELIMINARY (this article's reasoning, reading or arithmetic) | 21 |
| Labelled SPECULATIVE | 0 |
| Labelled UNVERIFIED | 0 |
The label distribution above is counted from the table below when this file is built. It is not typed by hand, which is the only way to keep it true.
What the labels mean
- VERIFIED — the claim is stated in a source that was retrieved and read during this investigation, and the source is named. Where the claim is a quotation, it was machine-checked against the retrieved text.
- PRELIMINARY — the claim is this article's own reasoning, interpretation or arithmetic, built on sources that are themselves verified. It is the part a reader is most entitled to argue with.
- SPECULATIVE — flagged as conjecture, with no retrieved source establishing it.
- UNVERIFIED — checked for and not found. Recorded as an absence, and never presented in the article as a fact about the world.
Sources
- Challenger, Gray & Christmas, Inc. — “October Layoffs at Three-Year Low; Seasonal Hiring Plans Up 26%”, The Challenger Report, published 5 November 2026. https://www.challengergray.com/blog/october-layoffs-at-three-year-low-seasonal-hiring-plans-up-26/ — retrieved in full; source for the 121,401 and 193,226 AI-attributed figures, the 617,189 year-to-date total, the 43,994 October total, the Market/Economic Conditions overtaking, the October AI figure of 1,265, the Microsoft voluntary exit offers, the Amazon and UPS seasonal announcements, the announced-hiring totals, and the Andy Challenger quotation.
- Challenger, Gray & Christmas, Inc. — “Challenger Report: August Job Cuts Up 58%, Consumer Products, Food Lead”, published 2 September 2026. https://www.challengergray.com/blog/challenger-report-august-job-cuts-up-58-consumer-products-food-lead/ — retrieved in full; source for the August AI figure of 3,462, the end of the five-month AI run beginning in March, the 116,175 year-to-date figure at approximately 22 per cent, the Restructuring lead, and the Andy Challenger quotation on hiring plans not being filled quickly.
- Challenger, Gray & Christmas, Inc. — “Challenger Report: Layoffs Fall, Hiring Picks Up; AI Leads For Fifth Straight Month”, published 6 August 2026. https://www.challengergray.com/blog/challenger-report-layoffs-fall-hiring-picks-up-ai-leads-for-fifth-straight-month/ — retrieved in full; source for the July AI figure of 10,970 at 33 per cent, the 112,713 figure at approximately 24 per cent, the fifth-consecutive-month statement, the Visa categorisation, the Montefiore case, the “Technological Update (possibly AI)” category and its 20,219 total for 2025.
- Brynjolfsson, E., Chandar, B. and Chen, R. — “Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence”, Stanford Digital Economy Lab, working paper, revised 12 August 2026. https://digitaleconomy.stanford.edu/publications/canaries-in-the-coal-mine/ — retrieved in full (abstract and authors' note); source for the 19 per cent gap for workers aged 22–25, the ADP payroll sample, the hiring-not-separations mechanism, the substitution-versus-complementation split, the compensation finding, all stated limitations, and the descriptive-not-causal framing.
- World Economic Forum — “The Future of Jobs Report 2025”, published 7 January 2025. https://www.weforum.org/publications/the-future-of-jobs-report-2025/ — retrieved in full (report summary page); source for the 170 million, 92 million and 78 million figures, the 22 per cent structural-transformation framing, the survey base of over 1,000 employers across 22 industry clusters and 55 economies, the five named drivers, and the fastest-growing roles.
- Gmyrek, P., Berg, J., Kamiński, K., Konopczyński, F., Ładna, A., Nafradi, B., Rosłaniec, K. and Troszyński, M. — “Generative AI and Jobs: A Refined Global Index of Occupational Exposure”, ILO Working Paper 140, 20 May 2025. https://www.ilo.org/publications/generative-ai-and-jobs-refined-global-index-occupational-exposure — retrieved in full; source for the one-in-four exposure figure, the 3.3 per cent highest-gradient figure, the female 4.7 per cent versus male 2.4 per cent split, the 11 per cent versus 34 per cent income split, the clerical finding, the task-survey methodology, and the transformation conclusion.
- International Labour Organization — “Generative AI and jobs: A 2025 update”, research brief, 20 May 2025. https://www.ilo.org/publications/generative-ai-and-jobs-2025-update — retrieved in full; source for the mean automation score of 0.29 versus 0.30, the standard deviation of 0.14 versus 0.30, the media and web task increase, and the “transformed rather than made redundant” statement.
- Statistics Canada — “Labour Force Survey, October 2026”, The Daily, released 6 November 2026. https://www150.statcan.gc.ca/n1/daily-quotidien/261106/dq261106a-eng.htm — retrieved in full; source for the October employment, employment rate, unemployment rate, the youth unemployment rate of 13.7 per cent, the September and August declines, wages, the household financial difficulty share, and job satisfaction.
- Statistics Canada — Table 14-10-0287-01, “Labour force characteristics, monthly, seasonally adjusted and trend-cycle”, DOI https://doi.org/10.25318/1410028701-eng, released 2026-11-06 — retrieved; the data table underlying source 8, cited so the series is checkable.
- U.S. Bureau of Labor Statistics — The Employment Situation, October 2026, USDL-26-1632, released 6 November 2026. https://www.bls.gov/news.release/empsit.nr0.htm — retrieved in full; source for the 165,000 payroll gain, the 4.1 per cent unemployment rate, the 4.1-to-4.3 per cent range since March, the August and September revisions, the average hourly earnings figure, and the sector employment statement.
- U.S. Bureau of Labor Statistics — Table A-1: Employment status of the civilian population by sex and age, October 2026. https://www.bls.gov/news.release/empsit.t01.htm — retrieved; source for the seasonally adjusted employment levels and participation rates from June to October 2026.
- U.S. Bureau of Labor Statistics — Job Openings and Labor Turnover, September 2026, USDL-26-1630, released 3 November 2026. https://www.bls.gov/news.release/jolts.nr0.htm — retrieved in full; source for 7.5 million openings, 5.4 million hires, 5.3 million separations, 3.4 million quits and 1.6 million layoffs and discharges.
- U.S. Bureau of Labor Statistics — JOLTS Table 2: Hires levels and rates by industry and region, seasonally adjusted, September 2026. https://www.bls.gov/news.release/jolts.t02.htm — retrieved; source for the Information sector hires of 94,000 and the sector hires levels used in section 9.
- Acemoglu, D. — “The Simple Macroeconomics of AI”, NBER Working Paper 32487, May 2024, prepared for Economic Policy. https://www.nber.org/papers/w32487 — retrieved in full (abstract); source for the 0.66 per cent and sub-0.53 per cent TFP estimates, the inequality findings, and the capital-labour conclusion.
- Autor, D. — “Applying AI to Rebuild Middle Class Jobs”, NBER Working Paper 32140, February 2024. https://www.nber.org/papers/w32140 — retrieved in full (abstract); source for the expertise-extension argument and the explicit statement that it is not a forecast.
- Bessen, J. — “Automation and jobs: When technology boosts employment”, VoxEU/CEPR column, 12 September 2019, summarising Bessen, “Automation and Jobs: When Technology Boosts Employment”, Economic Policy. https://cepr.org/voxeu/columns/automation-and-jobs-when-technology-boosts-employment — retrieved in full; source for the price-elasticity mechanism, the textile, steel and automobile employment history, the demand-satiation reading, and the reallocation-rather-than-mass-unemployment conclusion.
- Acemoglu, D. and Restrepo, P. — “Automation and New Tasks: How Technology Displaces and Reinstates Labor”, Journal of Economic Perspectives, vol. 33, no. 2, Spring 2019, pp. 3-30. https://www.aeaweb.org/articles?id=10.1257/jep.33.2.3 — retrieved (abstract and article information); source for the displacement and reinstatement effects, the new-tasks channel, and the finding that slower employment growth reflects a weaker reinstatement effect.
Claim-by-claim register
| # | Claim | Label | Source |
|---|---|---|---|
| 1 | Challenger reports AI cited in 121,401 job cut announcements in 2026 through October, approximately 20 per cent of all cuts | VERIFIED | 1 (direct quotation) |
| 2 | AI has been cited in 193,226 job cuts since 2023 | VERIFIED | 1 (direct quotation) |
| 3 | 121,401 divided by 617,189 is 19.7 per cent, consistent with Challenger's “approximately 20%” | PRELIMINARY (arithmetic) | 1; this article's arithmetic |
| 4 | 112,713 of 477,033 is 23.6 per cent, consistent with Challenger's July “approximately 24%” | PRELIMINARY (arithmetic) | 3; this article's arithmetic |
| 5 | 71,825 cuts were attributed to AI across 2023, 2024 and 2025 combined | PRELIMINARY (arithmetic) | 1 (193,226 minus 121,401); this article's arithmetic |
| 6 | Through October 2026 employers announced 617,189 job cuts, down 44 per cent from 1,099,500 in the same period of 2025 | VERIFIED | 1 (direct quotation) |
| 7 | October's 43,994 announced cuts were the lowest for the month since 2023 and down 71 per cent from 153,074 a year earlier | VERIFIED | 1 (direct quotation) |
| 8 | AI was the leading reason for announced cuts for five consecutive months, beginning in March 2026 | VERIFIED | 2, 3 (direct quotation of both) |
| 9 | AI led in July 2026 with 10,970 cuts, or 33 per cent | VERIFIED | 3 (direct quotation) |
| 10 | AI fell to fourth in August 2026 with 3,462 cuts, its lowest monthly total since December 2025 | VERIFIED | 2 (direct quotation) |
| 11 | Restructuring led in August 2026 with 16,173 cuts | VERIFIED | 2 (direct quotation) |
| 12 | Market/Economic Conditions led in October for the second consecutive month with 16,407 cuts, and has overtaken AI year-to-date at 130,531 | VERIFIED | 1 (direct quotation) |
| 13 | AI was cited for 1,265 cuts in October 2026, its lowest monthly total of the year | VERIFIED | 1 (direct quotation) |
| 14 | The fall from 10,970 in July to 1,265 in October is 88 per cent | PRELIMINARY | this article's calculation from 1 and 3 |
| 15 | Challenger's own reading is that reasons shifted from technology to the economy, and that the overall pace of cutting remains low | VERIFIED | 1 (direct quotation, Andy Challenger) |
| 16 | Challenger counts announced cuts by U.S.-based employers, categorised by the reason the company stated publicly | PRELIMINARY | this article's description of the series, from 1, 2 and 3 |
| 17 | Microsoft announced voluntary exit offers to 60,000 employees in October, not included in October's total | VERIFIED | 1 (direct quotation) |
| 18 | Technology announced 171,501 cuts through October, up 21 per cent, 28 per cent of all cuts | VERIFIED | 1 (direct quotation) |
| 19 | Technology's October total of 5,576 was its lowest monthly total of the year | VERIFIED | 1 (direct quotation) |
| 20 | Media announced 5,559 cuts through October, down 67 per cent from 16,680 | VERIFIED | 1 (direct quotation) |
| 21 | News, a subset of Media, announced 2,019 cuts through October, down 3 per cent | VERIFIED | 1 (direct quotation) |
| 22 | Health Care/Products announced 37,737 cuts through October | VERIFIED | 1 |
| 23 | Employers announced plans to hire 444,692 in October, the highest October total since at least 2009 | VERIFIED | 1 (direct quotation) |
| 24 | Year-to-date announced hiring plans were 655,304, up 34 per cent | VERIFIED | 1 (direct quotation) |
| 25 | Seasonal positions account for 72 per cent of all hiring plans announced in 2026 | VERIFIED | 1 (direct quotation) |
| 26 | Amazon announced 300,000 seasonal hires and UPS 100,000 | VERIFIED | 1 (direct quotation) |
| 27 | Challenger states that announced hiring plans do not appear to be filled quickly | VERIFIED | 2 (direct quotation) |
| 28 | Announced hiring plans cannot be added to announced cuts to produce a net employment figure | PRELIMINARY | this article's judgement from 1, 2 and 27 |
| 29 | Visa announced a 7 per cent reduction attributed to an efficiency push in which AI would reshape work, categorised by Challenger as AI | VERIFIED | 3 (direct quotation) |
| 30 | Montefiore eliminated 12 utilization review nursing positions after adopting Datavant software | VERIFIED | 3 (direct quotation) |
| 31 | A union characterised the Montefiore cuts as AI replacing workers; hospital leaders called that characterisation misleading | VERIFIED | 3 (reporting Challenger's account of the dispute) |
| 32 | Challenger categorised the Montefiore cuts as “Technological Update (possibly AI)” rather than AI | VERIFIED | 3 (direct quotation) |
| 33 | Challenger tracked 20,219 cuts in 2025 under the “Technological Update (possibly AI)” category | VERIFIED | 3 (direct quotation) |
| 34 | Those cuts are not included in the AI-attributed total | PRELIMINARY | this article's reading of 3 |
| 35 | The AI-attributed series records what companies stated publicly, and naming AI costs a company nothing while admitting weak demand does | PRELIMINARY | this article's analysis |
| 36 | Total nonfarm payroll employment rose 165,000 in October 2026 and unemployment was 4.1 per cent | VERIFIED | 10 (direct quotation) |
| 37 | The unemployment rate has remained between 4.1 and 4.3 per cent since March 2026 | VERIFIED | 10 (direct quotation) |
| 38 | August payrolls were revised from +133,000 to +186,000 and September from +29,000 to +60,000 | VERIFIED | 10 (direct quotation) |
| 39 | Average hourly earnings rose to $38.00, up 3.1 per cent over twelve months | VERIFIED | 10 (direct quotation) |
| 40 | Seasonally adjusted employment rose from 162,264 thousand in June 2026 to 163,615 thousand in October | VERIFIED | 11 (table data) |
| 41 | The participation rate rose from 61.5 to 61.9 per cent between June and October 2026 | VERIFIED | 11 (table data) |
| 42 | In September 2026 there were 5.4 million hires, 5.3 million total separations, 3.4 million quits and 1.6 million layoffs and discharges | VERIFIED | 12 (direct quotation) |
| 43 | Job openings were 7.5 million in September 2026 | VERIFIED | 12 (direct quotation) |
| 44 | Employers hired roughly 3.4 times as many people as they laid off or discharged in September 2026 | PRELIMINARY (arithmetic) | 12; this article's arithmetic |
| 45 | Information sector hires were 94,000 in September 2026, up from 70,000 in August, 62,000 in July and 71,000 in September 2025 | VERIFIED | 13 (table data) |
| 46 | Information sector hires rate was 3.4 per cent in September 2026, up 0.9 points | VERIFIED | 13 (table data) |
| 47 | Professional and business services hired 927,000 in September 2026 at a 4.1 per cent hires rate | VERIFIED | 13 (table data) |
| 48 | Health care and private education hired 798,000 in September 2026 at a 2.8 per cent rate | VERIFIED | 13 (table data) |
| 49 | Employment increased in professional and business services, health care, and transportation and warehousing in October | VERIFIED | 10 (direct quotation) |
| 50 | Challenger's 43,994 announced October cuts and JOLTS's 1.6 million September layoffs measure different things and different months | PRELIMINARY | this article's methodological judgement from 1 and 12 |
| 51 | Employment of workers aged 22–25 in AI-exposed occupations is 19 per cent below its peer-matched counterfactual | VERIFIED | 4 (direct quotation) |
| 52 | The Stanford analysis uses ADP administrative payroll data covering millions of U.S. workers through June 2026 | VERIFIED | 4 (direct quotation) |
| 53 | The gap operates primarily through reduced hiring rather than increased separations | VERIFIED | 4 (direct quotation) |
| 54 | Declines are concentrated where AI substitutes for human tasks; where it complements, employment is flat or rising | VERIFIED | 4 (direct quotation) |
| 55 | Adjustment is occurring through employment rather than base compensation | VERIFIED | 4 (direct quotation) |
| 56 | The authors describe the findings as early, descriptive indicators rather than causal estimates | VERIFIED | 4 (direct quotation) |
| 57 | The patterns attenuate when controlling for education, show trends predating generative AI, and are more pronounced in the ADP sample than in national surveys | VERIFIED | 4 (direct quotation) |
| 58 | The divergence has widened steadily since first documented in August 2025 | VERIFIED | 4 (direct quotation) |
| 59 | A 19 per cent gap is a comparison against a counterfactual, not a 19 per cent job loss | PRELIMINARY | this article's reading of 51 |
| 60 | No official statistical series records layoffs attributed to AI, nor entry-level hires by AI exposure | PRELIMINARY | absence across 10, 11, 12 and 13 |
| 61 | WEF projects 170 million jobs created, 92 million displaced and 78 million net growth between 2025 and 2030 | VERIFIED | 5 (direct quotation) |
| 62 | The WEF figures are extrapolations from survey respondents' predictions about structural labour-market transformation, not AI-only figures | VERIFIED | 5 (direct quotation of the framing sentence) |
| 63 | The WEF survey covers over 1,000 employers representing more than 14 million workers across 22 industry clusters and 55 economies | VERIFIED | 5 (direct quotation) |
| 64 | The WEF names five drivers, of which technological change is one alongside geoeconomic fragmentation, economic uncertainty, demographic shifts and the green transition | VERIFIED | 5 (direct quotation) |
| 65 | The three WEF percentages imply slightly different employment bases of roughly 1.11 to 1.21 billion | PRELIMINARY | this article's arithmetic from 61; consistent with rounding |
| 66 | WEF lists Big Data Specialists, Fintech Engineers, AI and Machine Learning Specialists and Software and Application Developers among the fastest-growing roles | VERIFIED | 5 (direct quotation) |
| 67 | Globally one in four workers are in an occupation with some GenAI exposure | VERIFIED | 6 (direct quotation) |
| 68 | 3.3 per cent of global employment is in the highest exposure gradient | VERIFIED | 6 (direct quotation) |
| 69 | Within that gradient, female employment is 4.7 per cent and male employment 2.4 per cent globally | VERIFIED | 6 (direct quotation) |
| 70 | Exposure is 11 per cent of total employment in low-income countries versus 34 per cent in high-income countries | VERIFIED | 6 (direct quotation) |
| 71 | In high-income countries the top gradient is 9.6 per cent female versus 3.5 per cent male | VERIFIED | 6 (direct quotation) |
| 72 | Clerical occupations continue to have the highest exposure levels | VERIFIED | 6 (direct quotation) |
| 73 | The ILO concludes that transformation of jobs is the most likely impact of GenAI | VERIFIED | 6 (direct quotation) |
| 74 | The ILO index was built from 29,753 tasks, a survey of 1,640 people, 52,558 data points and expert review | VERIFIED | 6 (direct quotation) |
| 75 | The 2025 mean automation score is 0.29 versus 0.30 in 2023, with standard deviation 0.14 versus 0.30 | VERIFIED | 7 (direct quotation) |
| 76 | The ILO states that most jobs will be transformed rather than made redundant | VERIFIED | 7 (direct quotation) |
| 77 | Growing GenAI capability in voice, image and video has raised automation scores for media- and web-related tasks | VERIFIED | 7 (direct quotation) |
| 78 | Exposure measures the potential for task automation, not observed job loss | PRELIMINARY | this article's reading of 6 and 7 |
| 79 | StatCan reported October 2026 employment little changed at 21.1 million, employment rate 60.5 per cent, unemployment rate 6.7 per cent | VERIFIED | 8 (direct quotation and table data) |
| 80 | Canadian youth unemployment for ages 15 to 24 rose 0.7 points to 13.7 per cent | VERIFIED | 8 (direct quotation) |
| 81 | Canadian employment declined in August and September 2026 | VERIFIED | 8 |
| 82 | Canadian average hourly wages rose 2.2 per cent year over year to $37.89 | VERIFIED | 8 (direct quotation) |
| 83 | 29.7 per cent of Canadians aged 15 and older lived in a household finding it difficult to meet financial needs, up 2.0 points in twelve months | VERIFIED | 8 (direct quotation) |
| 84 | 60.9 per cent of employed Canadians reported being very satisfied with their job | VERIFIED | 8 (direct quotation) |
| 85 | No retrieved source connects the October 2026 Canadian youth unemployment movement to AI adoption | PRELIMINARY | absence across 8 and 9 |
| 86 | Canada's youth unemployment rate is about twice its national rate (13.7 per cent against 6.7 per cent) | PRELIMINARY (arithmetic) | 8; this article's arithmetic. An earlier draft said 'more than three times', which 13.7 divided by 6.7 does not support. |
| 87 | Autor argues AI's opportunity is to extend the relevance, reach and value of human expertise | VERIFIED | 15 (direct quotation) |
| 88 | Autor states his thesis is not a forecast but an argument about what is possible | VERIFIED | 15 (direct quotation) |
| 89 | Acemoglu estimates AI could raise TFP by no more than 0.66 per cent over ten years, and predicts less than 0.53 per cent | VERIFIED | 14 (direct quotation) |
| 90 | Acemoglu finds no evidence AI will reduce labour income inequality, and predicts a widening gap between capital and labour income | VERIFIED | 14 (direct quotation) |
| 91 | Autor's complementarity argument and the Stanford substitution finding describe opposite deployment choices of the same technology | PRELIMINARY | this article's reading of 14, 15 and 4 |
| 92 | No net AI employment figure is supportable from available data | PRELIMINARY | this article's conclusion from 1-15 |
| 93 | September 2026 payrolls were first published at +29,000 and revised to +60,000 within a month | VERIFIED | 10 (direct quotation) |
| 94 | Professional and business services hired 957,000 in August 2026, down to 927,000 in September | VERIFIED | 13 (table data) |
| 95 | 43,994 announced October cuts are 2.75 per cent of the 1.6 million layoffs and discharges recorded in September 2026 | PRELIMINARY (arithmetic across two different measures and months) | 1, 12; this article's arithmetic |
| 96 | AI-attributed cuts fell from about 24 per cent of all announced cuts, year-to-date through July, to about 20 per cent year-to-date through October, so the share fell even as the absolute cumulative figure set a record | PRELIMINARY (arithmetic) | 1, 3; this article's arithmetic |
| 97 | The 121,401 AI-attributed cuts in 2026 are 62.8 per cent of the 193,226 recorded since 2023 | PRELIMINARY (arithmetic) | 1; this article's arithmetic |
| 98 | In a competitive market, cutting the labour needed per unit of output reduces the price, and if demand is sufficiently elastic, demand rises fast enough that employment increases even as labour per unit declines | VERIFIED | 16 (direct quotation) |
| 99 | US employment in cotton textiles, primary steel and automobiles grew alongside rapid productivity growth for a century or more before falling | VERIFIED | 16 |
| 100 | Acemoglu and Restrepo: automation has a displacement effect that may reduce labor demand even as it raises productivity, counterbalanced by a reinstatement effect from new tasks that always raises the labor share and labor demand | VERIFIED | 17 (direct quotation) |
| 101 | Their decomposition attributes slower US employment growth over the last three decades to faster displacement, a weaker reinstatement effect and slower productivity growth | VERIFIED | 17 |
| 102 | Some industries today with substantial unmet demand will respond elastically to automation and are likely to see rising employment, so the elasticity of demand determines the sign of the change | VERIFIED | 16 (direct quotation) |
| 103 | Recent studies find positive employment responses to information technology, automation and robotics in some industries, no effect on employment in others, and a negative effect in one | VERIFIED | 16 (direct quotation) |
| 104 | Workers must adapt to new industries, skills and occupations as automation relocates jobs between industries | VERIFIED | 16 (direct quotation) |
Stated limitations and open questions
These are carried in the article's own outstanding list. They were not closed by this investigation, and they are the first things a critic should press on.
- No interview was conducted. No economist, statistician, policy analyst or worker was contacted. Every position here is a published position, in the authors' own words. The two most important sources — the Challenger series and the Stanford paper — are both interpreted without access to their authors. A twenty-minute conversation with Andy Challenger would have settled the announcement-versus-event question better than the article's own reasoning does, and did not happen.
- The September 2026 monthly AI figure is not in the record. Challenger's October release gives 121,401 year-to-date and 1,265 for October; the August release gives 116,175. September's own monthly attribution is therefore only known by subtraction, and this article does not present it. Anyone repeating a September figure should find the September release, which was not retrieved.
- JOLTS and Challenger cover different universes and different months. Challenger is announcements by U.S.-based employers; JOLTS is payroll turnover in the establishment survey. This article uses JOLTS to establish the size of the flows and Challenger to establish what companies said. It does not and cannot reconcile them, and no one has.
- The Stanford 19 per cent figure is a working paper, not a peer-reviewed publication. It is revised, dated 12 August 2026, and widely cited, but it has not been through journal review in the version retrieved. Its limitations are stated by its authors and quoted at length in section 5.
- The WEF figures are from the 2025 report and are approaching the end of their 2025-2030 window without a newer edition retrieved. A 2027 edition would change the arithmetic and the article would need revision.
- No company was contacted about its own layoff attribution. Visa and Montefiore appear only as Challenger described them. The hospital dispute is presented as Challenger presented it — a union position and a hospital position — with no independent reporting from this desk.
- The Canadian section is one month of one survey. Statistics Canada's October release is a snapshot, several of its series are seasonally adjusted with revision risk, and no Canadian study equivalent to the Stanford paper was located. The section says so.
- No sector-level series exists for customer support or entry-level hiring. Section 9 says the evidence is thin in those categories because it is. Absence of measurement is recorded as absence, not as evidence of absence.
- Nine passages were converted from verbatim quotation to reported speech during verification. The house quotation verifier tests each quoted span against the retrieved source text and does not normalise whitespace across the source's own hard line breaks. Several long spans in the first draft therefore failed even though their wording was exact, because the government releases wrap a sentence mid-clause. Rather than trim them into fragments and keep quotation marks the verifier could not confirm, those nine passages now report the figures in this article's own voice with the same [VERIFIED] citation. The claim is unchanged; only the pretence of verbatim quotation is removed. The spans that remain in quotation marks are the ones the verifier tested and found.
- Two house-template strings contradict this package and were not changed here. The FOR HUMAN CONSUMPTION panel carries a standing note stating that the synopsis introduces no source, which is untrue of this synopsis: it names Challenger, Gray & Christmas, Stanford and the World Economic Forum. The evidence page footer carries the slogan 'Human approved. Mostly.' on a page that in the same file states the article must not be published and that no approval has been given. Both strings come from shared tooling rather than from this article, and both would be false on any of the staged pages, so they are recorded as a house-wide defect to fix in the templates rather than patched on one article. An independent reviewer flagged both on 2026-11-10.
What this file does not claim
It does not claim that every VERIFIED line is beyond dispute — verification here means the source says it, not that the source is right. It does not claim that an absence is a finding: where a claim is labelled UNVERIFIED, the honest reading is that this desk looked and did not find it, not that it is not so. And it does not claim to be exhaustive in the other direction either: a few figures appear inside a directly quoted source span without a row of their own, and one row (Microsoft's voluntary exit offers) is background the article does not use.