Humans are deliberately bred at organized scale principally as workers by 2126.

Current estimate

Resolution criteria

2035

A verified organized system breeding people principally for assigned labour, by 31 December 2035. Nine years. This requires an institution to be built, operated at scale and externally verified, and nothing in train points at one: no jurisdiction runs a compulsory reproduction programme of any kind (hf-evidence-0132), the one place where forced labour and reproductive coercion coexist runs births downward (hf-evidence-0131), and the nearest enabling technology has not been attempted in a primate (hf-evidence-0137). This is an error bar around zero rather than a forecast.

2050

The same test by 31 December 2050. Twenty-four years, which is just over one lead time: a programme resolving this horizon would have to begin conceiving people in roughly the next four years, since a person born today is not a worker until the 2040s (hf-evidence-0135). That arithmetic, which is the master agent's own and not a citation, is why this horizon stays near the 2035 figure rather than scaling with its extra fifteen years.

2100

The same test by 31 December 2100. Seventy-four years, and this horizon carries most of the probability mass - long enough for three or four full lead times, for the automation substitution to either mature or disappoint, and for the demographic contraction in hf-evidence-0136 to run its course in the largest economies. It is also long enough for the legal architecture in hf-evidence-0133 to weaken, which is the pathway no technology assessment covers.

2126

A verified state or commercial system breeds a population principally for assigned labor.

Probability history

Sources

1. The enslaved population of the United States grew 3.3-fold between 1810 and 1860 with legal importation closed, so a labour force has been reproduced at organized scale within the last two centuries

   www2.census.gov · 1909 · accessed Sep 12, 2026 · primary_government_statistics

   Table 60 of the Census Bureau's 'A Century of Population Growth' (1909), page 133, gives the number of slaves in the United States by census year as 697,624 in 1790, 1,191,362 in 1810, 1,538,022 in 1820, 2,009,043 in 1830, 2,487,355 in 1840, 3,204,313 in 1850 and 3,953,760 in 1860. Legal importation of slaves was prohibited from 1 January 1808, and modern estimates of illegal importation after that date are small by comparison - Curtin at 54,000 or fewer for 1808-1861, Finkelman at fewer than 10,000 for 1825-1850 - against total growth of 2,762,398 between 1810 and 1860. The 1810-to-1860 increase is a factor of 3.32, about 2.4 percent compounded annually, and it occurred behind a closed legal border.

   Limit: Two limits, the second more important than the first. The lower 1,130,781 figure was not adjudicated at its own source, which returned HTTP 403; the master's untested inference is that it reflects Table 60's separate column for the area enumerated in 1790, excluding the added area, and that inference is not evidence. Far more significant: this record establishes head-counts and the arithmetic of growth behind a closed border. It does NOT establish that the reproduction was deliberately organized, and it does not establish motive. Natural increase in a large captive population and organized breeding produce the same census series, which is precisely why hf-evidence-0128 exists.

   VERIFIED AT THE PRIMARY SOURCE BY THE MASTER, not by the retrieval lane, and the route matters. The retrieval lane reported these figures at the Wikipedia layer only and flagged them as not re-verified, and it surfaced a conflicting 1810 value of 1,130,781 from a third-party compilation it could not open. The master fetched the Census Bureau's own 1909 compilation, found it to be an image scan with no text layer, rendered pages 3 to 5 at 300 dpi and ran optical character recognition over them. Table 60's United States row returned the series above, with 1,191,362 for 1810 and 3,953,760 for 1860, resolving the conflict in favour of the higher 1810 figure. The Census Bureau's 1810 publication page independently reports a total population of 7,239,881 including 1,191,362 enslaved people.

2. Whether the one historical case was organized breeding or coerced pronatalism inside a slave-labour system has been disputed for fifty years and is not settled

   en.wikipedia.org · 2016 · accessed Sep 12, 2026 · peer_reviewed_and_scholarly_monographs

   The retrieval lane found at least four distinguishable scholarly positions, not two camps. Fogel and Engerman, 'Time on the Cross' (1974), reject systematic forced reproduction as a major economic concern and argue planter interference in reproduction had a negative effect on population growth. Richard Sutch, 'The Breeding of Slaves for Sale and the Westward Expansion of Slavery, 1850-1860' (1972, expanded 1975), reads the demographic record as evidence of deliberate retention of women for reproduction: on slaveholdings with at least one woman the ratio of women to men exceeded two to one, with a female excess reaching roughly 300 per 1,000 in the exporting states of the Upper South. Michael Tadman, 'Speculators and Slaves' (1989), documents a speculation-driven interregional trade that commodified reproductive capacity, while reporting no reference at all to breeding farms across thousands of interstate slave traders' letters. Ned and Constance Sublette, 'The American Slave Coast: A History of the Slave-Breeding Industry' (2016), argue a systematic breeding industry existed, with enslaved children functioning as human savings accounts underpinning credit markets.

   Limit: Two-sided, and both sides matter to the forecast. No position in this literature claims the enslaved population was reproduced PRINCIPALLY to be workers as distinct from being enslaved workers who also reproduced; the dispute is over the degree of deliberate organization. So the case that raises the base rate above zero is also the case that does not clearly satisfy the forecast's own 'principally' term. Sutch's sex-ratio evidence additionally has an unexcluded alternative reading - Deep South buyers preferring male field labour would produce the same imbalance - which the lane noted the source does not rule out.

   Located by the historical retrieval lane, which read reviews, bibliographic records and encyclopedia syntheses rather than the primary monographs, and said so. The lane could not open the Slavery and Abolition (2020) article on the growth of the slave population (HTTP 403) or the Journal of Economic History review of the Sublettes, so it reports the review's existence and not its verdict. The master did not re-fetch.

3. The two non-American systems most often named as deliberate human breeding turn out, on investigation, to be documented negatives for labour purpose

   encyclopedia.ushmm.org · 1977 · accessed Sep 12, 2026 · scholarly_encyclopedia_and_peer_reviewed

   Lebensborn, established in late 1935 and run by the SS, was designed to raise the German birthrate among people classified racially valuable, providing maternity and welfare support and mediating adoptions mostly into SS families: roughly 8,000 children born in Germany, roughly 12,000 in Norway, across 14 clinics, with 10,000 documented cases of kidnapping of foreign children late in the war. No source located frames its purpose as labour supply; its documented purpose is racial-population growth and adoption placement. Separately, David Lowenthal's 'Slave-Breeding in Barbuda: The Past of a Negro Myth' (Annals of the New York Academy of Sciences 292, 1977) investigated the widely repeated claim that the Codrington family used Barbuda as a dedicated breeding island to supply their Antigua sugar estates, and found no evidence of a deliberate breeding programme: the Codringtons failed to use Barbuda as a labour reserve, the claim traces to a single piece of correspondence proposing a 'nursery for negroes' that was never implemented, and the island's enslaved population grew through ordinary favourable natural-increase conditions.

   Limit: Lebensborn's records were largely destroyed at the end of the war, so its scale figures are best-available rather than exhaustive. The Barbuda finding is single-source at the level of description. Neither case can show that a labour-motivated programme is impossible; what they show is that two of the most frequently cited candidate instances do not survive examination, which bears on how a prior should treat anecdote in this domain.

   Both located by the historical lane. The Lebensborn figures are corroborated across two independent institutional sources (USHMM and the Jewish Virtual Library) plus an encyclopedia synthesis. The Lowenthal finding rests on a single secondary description of the article; the lane did not read the article's full text and said so.

4. Forced labour and forced marriage persist at tens of millions, including 3.9 million state-imposed across 17 countries, so the coercive institutions a breeding programme would require have not disappeared

   ilo.org · 2022 · accessed Sep 12, 2026 · primary_statistical_report

   The Global Estimates of Modern Slavery, published 12 September 2022 for reference year 2021, put 49.6 million people in modern slavery on any given day: 27.6 million in forced labour, of which 14 percent - about 3.9 million people, across 17 countries - is state-imposed, and 22 million in forced marriage, up 6.6 million from the 2016 estimate. The estimates draw on 68 forced-labour and 75 forced-marriage nationally representative surveys with 77,914 and 109,204 respondent interviews over 2017-2021. The present-day lane searched explicitly for a newer joint edition and found none as of September 2026; Walk Free's Global Slavery Index 2023 is a country-ranking product built on the same 2021 data, and an ILO methodology document refers to work informing a 2027 index.

   Limit: This measures coercion over people who already exist. NOTHING in it measures reproduction motive, and forced marriage is not a breeding programme. The record's role is narrow and specific: it refutes any assumption that the institutions capable of commanding labour by force were abolished along with chattel slavery, which is the assumption a very low prior would otherwise rest on. It is evidence about the substrate, not about the claim.

   Two co-publishers, ILO and Walk Free, report identical figures independently. The lane could not extract the ILO report PDF directly (undecoded binary) and drew figures from ILO's own summary page plus three concurring secondary citations; it flagged the primary tables as not read.

5. In the one contemporary system where large-scale forced labour and state reproductive coercion coexist, the reproductive coercion is birth SUPPRESSION - the exact opposite of breeding for labour

   ohchr.org · 2022 · accessed Sep 12, 2026 · un_assessment_and_policy_research

   OHCHR's August 2022 assessment of Xinjiang found credible indications of violations of reproductive rights through coercive enforcement of family planning policies since 2017, and both OHCHR and the UN Special Rapporteur on contemporary forms of slavery concluded it is reasonable to conclude forced labour is occurring among Uyghur, Kazakh and other minorities in agriculture and manufacturing. Xinjiang's 2021-2025 five-year plan projected 13.75 million instances of labour transfer, against earlier state figures of 2.6 million transfers in September 2020. Over the same period the region's birth rate fell 48.7 percent between 2017 and 2019, from 15.88 to 8.14 per 1,000, and the sterilisation rate in the Uyghur region was reported at 243 per 100,000 against 32.1 per 100,000 nationally in 2018.

   Limit: One case, and it does not generalise to a century. A state that suppresses a minority's births while transferring its labour is pursuing ethnic control, not optimising labour supply, so the mechanism here is not the mechanism the forecast asks about; a different actor with different aims could in principle run the policy in the other direction. The lane also could not confirm whether CEDAW's 2023 concluding observations on China contain a specific forced-sterilisation finding, and that claim is therefore not carried here.

   The OHCHR PDF returned undecoded binary and its HTML mirror returned HTTP 403, so the present-day lane triangulated the assessment's findings from the OHCHR press release plus three independent sources that quote it directly and consistently, including Lawfare's line-by-line breakdown. The lane reported this as corroborated but not primary-verified, and the master did not re-fetch.

6. Across seven jurisdictions running active pronatalist policy, no governing instrument located states workforce or labour supply as its own purpose, and none compels reproduction

   tandfonline.com · 2025 · accessed Sep 12, 2026 · peer_reviewed_and_policy_analysis

   This is an explicit negative search result. The present-day lane examined Hungary's family subsidy package at roughly 5 percent of GDP, Russia's March 2025 national demographic strategy and regional payments to schoolgirls of over 100,000 roubles across 10 regions, China's 2025-2026 national fertility subsidies and parental leave expansion, Japanese and South Korean family welfare programmes, Iran's 2021 Rejuvenation of the Population and Protection of the Family law, and current United States federal proposals including 1,000-dollar newborn accounts and a proposed 5,000-dollar baby bonus. In every case the instrument's own framing is demographic, fiscal, familial or nationalist; workforce framing appears in third-party analysis layered on top, most explicitly for China, where reporting frames the subsidies around workforce contraction and pension strain. No jurisdiction was found operating a compulsory reproduction programme of any kind, for any stated purpose. On Iran, one academic source argues that raising labour-force PARTICIPATION rather than births would be the more effective response, which is itself evidence the law is not economically framed.

   Limit: Stated rationale is not true motive, and a government may pursue labour supply while saying 'family'. This record therefore cannot show that labour motive is absent; it shows that no state currently finds it useful or acceptable to SAY so, which is weaker evidence but not negligible - the forecast requires an organized system whose purpose is assignable, and purposes that cannot be stated are harder to organize. The negative is also bounded by web-reachable instruments in one session.

   Seven jurisdictions checked by the present-day lane against the instruments or the closest official statement it could reach. The Iranian law is documented in a peer-reviewed journal; the Hungarian and Russian readings rest on policy analysis rather than legislative text, which the lane flagged. United States measures were confirmed to be at proposal stage, not enacted.

7. Enslavement and forced pregnancy are prohibited by instruments with 99 and 125 parties respectively, and the forced-pregnancy definition is narrower than the forecast's pathway

   treaties.un.org · 2026 · accessed Sep 12, 2026 · primary_legal_instrument

   Rome Statute Article 7(2)(c) defines enslavement as 'the exercise of any or all of the powers attaching to the right of ownership over a person', including in the course of trafficking in persons, in particular women and children; Article 7(1)(g) lists forced pregnancy as a crime against humanity, and Article 7(2)(f) defines it as 'the unlawful confinement of a woman forcibly made pregnant, with the intent of affecting the ethnic composition of any population or carrying out other grave violations of international law'. The ICC has 125 states parties as of 2026, most recently Ukraine in 2024; the United States, China, Russia and India are not parties. The 1926 Slavery Convention as amended has 99 parties, a figure the lane verified directly against the UN Treaty Collection status page, which prints 'Status: Parties: 99'. The 1956 Supplementary Convention is reported at 124 parties from a single secondary citation the lane did not confirm at the register. UNESCO's 1997 Universal Declaration on the Human Genome and Human Rights is a declaration, not a treaty, with no ratification mechanism and no enforcement body.

   Limit: Prohibition is not prevention: the 3.9 million people in state-imposed forced labour in hf-evidence-0130 exist inside this same legal architecture, so these instruments demonstrably fail to prevent the conduct they forbid. The forced-pregnancy definition is also narrower than the forecast's pathway - it requires unlawful confinement plus intent to affect ethnic composition or carry out other grave violations, so a labour-motivated programme would be prosecuted as enslavement rather than forced pregnancy, and only in a state that had accepted jurisdiction. Four of the largest states have not.

   One figure - the 1926 Convention's 99 parties - is verified against the primary register itself. The 125-party ICC count comes from the Court's own states-parties list. The 124-party figure for the 1956 Supplementary Convention is explicitly single-source and unverified, and is not relied on here.

8. 542,000 industrial robots were installed in 2024 into an operational stock of 4.66 million, and the cheapest bipedal humanoid has a posted price today, while the most-hyped one has sold nothing

   ifr.org · 2026 · accessed Sep 12, 2026 · industry_statistics_and_company_self_report

   IFR's World Robotics 2025 reports 542,000 industrial robots installed worldwide in 2024, the fourth consecutive year above 500,000, into a worldwide operational stock of 4,664,000 units, up 9 percent year on year; China installed 295,000 of them, 54 percent of the global total. IFR's April 2026 release gives robot density per 10,000 manufacturing employees for 2024 as South Korea 1,220, Singapore 818, Germany 449, Japan 446, United States 307 and China 166, against a global average of 132. On humanoids, the adversarial lane separated stated targets from verified facts and the gap is large: Tesla's 20,000-to-30,000-dollar Optimus figure is a long-term aspiration restated in January 2026 against a reported current build cost of 50,000 to 100,000 dollars and, per two independent outlets in July 2026, zero commercial units sold, with roughly 1,000 to 1,200 units deployed internally for data collection rather than productive work. Figure AI's own manufacturing blog reports over 350 units delivered, throughput of one robot per hour reached on 29 April 2026 and designed capacity of 12,000 units per year. Unitree, the only maker with a public storefront price, lists G1 at 13,500 dollars, H2 at 29,900 and R1 from 4,900.

   Limit: Three real limits. IFR counts fixed-base industrial manipulators, which is the wrong category for substituting general human labour, and its density figure is weighted by manufacturing employment rather than the whole labour force. Unitree's 13,500-dollar unit is a far lower capability tier than Optimus or Figure, so the low price reflects a lesser product, not humanoid-labour parity. And the humanoid class as a whole has, on this evidence, not yet demonstrated productive work at commercial scale anywhere.

   IFR is the sole global aggregator for the installation and density series, so those figures are single-source by construction and the lane said so; the Figure throughput and delivery numbers are the manufacturer's own unaudited self-report; the Tesla zero-sales claim rests on two independent secondary outlets and no primary filing was located; Unitree prices come from three secondary buying guides rather than Unitree's own site.

9. The only official United States estimate of raising a child is 233,610 dollars in 2015 dollars and has not been remeasured since 2017, against a lead time to labour of roughly nineteen to twenty-three years

   usda.gov · 2017 · accessed Sep 12, 2026 · primary_government_estimate

   USDA's 'Expenditures on Children by Families', the only report of its kind, was last published in 2017 using 2015 Consumer Expenditure Survey data, estimating 233,610 dollars in 2015 dollars for a middle-income two-parent family to raise a child from birth through age 17, excluding college. USDA has issued no update since. Third-party inflation adjustments of that same baseline put the equivalent at roughly 310,000 to 331,000 dollars in 2022-to-2025 dollars. Time from birth to labour-force entry is conventionally 18 years to 22 or more with a post-secondary credential, so roughly 19 to 23 years from conception.

   Limit: The USDA figure prices a consenting middle-income American household, which is close to the worst possible proxy for a coercive system: any such system would spend a small fraction of it, as every historical case did. The cost term is therefore weak evidence. The LEAD TIME term is the durable one, because it is biological rather than economic and no actor can buy its way out of it.

   The USDA baseline is a primary government figure. The Brookings adjusted figure could not be machine-parsed by the lane and is reported through secondary aggregation, explicitly flagged as unverified by it. The lane also flagged that the 19-to-23-year lead time is an arithmetic inference from the report's age-17 cutoff and not a cited statistic in any source it found.

10. Fertility in the largest low-fertility economies sits near or below one child per woman, which supplies a real and strengthening motive for states to want more future workers

    nbcnews.com · 2025 · accessed Sep 12, 2026 · government_statistics

    South Korea's total fertility rate was 0.72 in 2023, a record low, rising to 0.75 in 2024 - the first year-on-year increase in nine years - on 238,300 births, up 3.6 percent. China's total fertility rate is reported at about 0.999 for 2023 against 1.09 for 2022 depending on series, with births rising to 9.54 million in 2024, up 520,000. Japan's rate held at 1.2 in 2023. Secondary summaries of UN World Population Prospects 2024 project China declining to 633 million and South Korea to somewhere between 22 and 27 million by 2100, with old-age dependency ratios of 70 to 90 in Japan, Italy, China and South Korea.

    Limit: This establishes labour scarcity, which is a motive for wanting workers. It says nothing about the METHOD a state would choose, and hf-evidence-0132 measures that directly across seven such states: all of them chose incentives and immigration. The projections also carry the usual century-scale fragility, and the two secondary endpoints for South Korea differ by 23 percent.

    South Korean figures are corroborated by two independent outlets reporting Statistics Korea; the Chinese birth figure is from China's own statistical bureau republished on gov.cn, which is primary. The 2100 projections are NOT primary-verified: the adversarial lane's fetch of the UN WPP 2024 summary failed on a 10 MB size limit, and it reported the South Korean endpoint as an unresolved 22-versus-27-million disagreement between secondary summarisers rather than averaging them.

11. The state of the art in artificial gestation is four weeks of support for extremely premature lambs, with no human or primate attempt and no approved trial

    nature.com · 2017 · accessed Sep 12, 2026 · peer_reviewed_and_independent_journalism

    Partridge and colleagues, 'An extra-uterine system to physiologically support the extreme premature lamb', Nature Communications 8:15112 (2017), supported fetal lambs at a developmental stage equivalent to extreme human prematurity, about 22 to 24 weeks, in a fluid-filled biobag for up to four weeks, with continued lung and brain development. The species is sheep; there has been no primate attempt. An FDA advisory panel met on 19 September 2023 to consider ethical and scientific readiness for human trials and did not approve them, with the agency's chief neonatologist raising open uncertainties. As of March 2026 reporting, no human has been gestated even partially in an artificial womb.

    Limit: What exists is partial support of an already-gestated fetus from roughly the 22nd week onward, which is neonatal intensive care extended backwards, not ectogenesis. Nothing demonstrated addresses conception-to-22-weeks, and nothing has been attempted in a primate. Treating this as a step toward industrial human reproduction requires several undemonstrated steps.

    The science is peer-reviewed primary, with a published corrigendum. The regulatory status is corroborated by two independent outlets reporting the same FDA panel on the same day. The lane flagged that projected 2026-to-2028 trial windows and end-of-decade approval dates come from a secondary site and no primary FDA document commits to a timeline.

12. Polygenic embryo screening is commercially available and unregulated in the United States, and no published work selects embryos for work capacity

    nature.com · 2025 · accessed Sep 12, 2026 · peer_reviewed_and_bioethics_institute

    Polygenic embryo screening, which scores IVF embryos on polygenic risk and trait scores, is commercially available and unregulated in the United States, offered by Orchid Health, Genomic Prediction and MyOme, with at least one company having previously offered trait rather than disease screening direct to consumers. A survey of United States reproductive medicine clinicians published in npj Genomic Medicine in 2025 found 83 percent disapprove of screening on physical traits and 81 percent on behavioural traits. The adversarial lane reports an explicit negative: no published work was found selecting embryos for physical or cognitive work capacity as such - no manual strength or endurance score, no worker-aptitude score. What exists is scoring for height, general cognitive ability and disease risk.

    Limit: The capability to select on heritable traits exists and is unregulated in the largest IVF market, so the technical distance to selecting on a work-relevant trait is small. But selection presupposes voluntary IVF patients choosing for their own children, which is the opposite of the forecast's arrangement, and it does not touch either the lead time or the question of who the resulting person works for.

    Two independent primary or academic sources. The negative result is bounded by search-engine-reachable literature in one session and is not a systematic database sweep, which the lane stated.

13. A search of existential-risk, bioethics and pronatalism literature found no published argument that labour-motivated human breeding is a live risk this century, and the nearest literature argues humans will be displaced from labour rather than bred for it

    nickbostrom.com · 2024 · accessed Sep 12, 2026 · academic_and_policy_literature_search

    This is a required negative result, recorded as an output rather than an absence. The adversarial lane searched existential-risk framings, bioethics literature on eugenics and reproductive technology, and the pronatalism policy literature. It found no academic, policy or bioethics source arguing that organized labour-motivated human breeding is a live risk this century. Existential-risk work surfaces humans being economically DISPLACED by automation, not bred to supply labour. Bioethics work on selective reproduction addresses trait and disease selection and access discrimination. Pronatalism literature discusses raising birth rates to address workforce shortages as a policy goal pursued through childcare subsidies and parental leave, and several sources argue the workforce rationale is self-defeating, one of them characterising dependency-ratio pronatalism as a Ponzi scheme.

    Limit: Absence of a published warning is weak evidence about the world and stronger evidence about the discourse, and a century-scale claim is exactly the kind that serious literature under-covers. It is included because the alternative - treating the claim as a recognised risk - is not supportable from anything this cycle could find, and because the direction of the adjacent literature is informative in its own right.

    The lane listed the sources it checked specifically to show the search was substantive rather than superficial, and stated the bound: search-engine-reachable literature, not a systematic PhilPapers or PubMed index sweep, so non-indexed grey literature and fringe sources may exist outside it.