Modern Slavery in the Twenty-First Century

A Long Read · Investigation

Resurgence, recognition, or statistical mirage?

By Kevin J. Parker

To hold another human being as property—to bend a life wholly to one’s will and call the arrangement ownership—is among the oldest of human crimes, and it was never, in any age or under any flag, anything other than a crime. No economy required it that could not have been built otherwise; no custom hallowed it that was not already a wound; no ignorance excused it, for the enslaved always knew, and said, precisely what was being done to them. If slavery is more abhorrent now, it is not because the act has darkened but because we have run out of alibis: we have abolished it in the law of every nation on earth, counted its victims, named its mechanisms, and acquired both the knowledge and the means to end it. That it survives is therefore no longer the inherited blindness of an older world but a decision to look away. The wrong was always absolute; what is new is our nakedness before it—and the fact that it now hides not in some distant field but in the ordinary machinery of the lives we call free.

There is a fair chance that the solar panel feeding clean electricity into a European or Australian rooftop began its life in a place its owner has never heard of and would rather not contemplate. Roughly nine in ten of the world’s solar modules depend on a single refined material, polysilicon, and for much of the past decade close to half the global supply of that material has come from one region: Xinjiang, in north-western China.¹ Researchers at Sheffield Hallam University traced the chain from quartz rock to finished panel and concluded that the four largest manufacturers in the world were exposed, at some tier of their supply, to a state-run programme of coerced labour imposed on Uyghurs and other minorities.² The technology of a low-carbon future, in other words, may rest on one of the oldest forms of human exploitation.

That uncomfortable entanglement is precisely why ‘modern slavery’ has migrated from the language of charities into the machinery of trade. The United States now blocks imports it suspects of being tainted by forced labour, and the volume of those seizures, the industries they touch and the diplomatic friction they generate have made the question urgent in a way it was not a decade ago. But urgency is not the same as clarity. Before asking what to do about modern slavery, a sceptic is obliged to ask a prior question: is it genuinely on the rise, or are we simply, after centuries of looking away, finally learning to see and to count what was always there?

ONE What we talk about when we talk about slavery

The first difficulty is definitional, and it is not trivial. ‘Modern slavery’ is not a category in international law. The United Nations itself describes it plainly as an umbrella term, one that gathers a cluster of distinct legal wrongs—forced labour, debt bondage, forced marriage, servitude and human trafficking—around a shared core: a situation a person cannot refuse or leave because of coercion, deception, violence or abuse of power.³ The phrase does rhetorical work that the underlying legal terms cannot. It is vivid, morally charged and politically mobilising. It is also imprecise.

The component concepts, by contrast, are defined with some care. The International Labour Organization’s Forced Labour Convention of 1930 (No. 29) frames forced labour as work exacted from a person under the menace of a penalty and for which that person has not offered themselves voluntarily—a formulation still in force, and one of the most widely ratified labour standards in existence.⁴ State-imposed forced labour is addressed separately in the Abolition of Forced Labour Convention of 1957 (No. 105). Human trafficking acquired its first binding international definition only in 2000, through the Palermo Protocol supplementing the UN Convention against Transnational Organized Crime, which fixed the now-familiar triad of act, means and purpose, and made a victim’s consent legally irrelevant once coercion or deception is present.⁵

Scholars have pressed harder objections than mere imprecision. Some argue that the slavery analogy imports the moral weight of the transatlantic trade onto situations that may not resemble chattel bondage, flattening morally distinct harms into a single emotive category and, in the process, shaping which victims attract resources and which are overlooked. Others observe that the term’s very elasticity serves its promoters: a problem defined expansively can be declared vast, and a problem declared vast attracts funding, legislation and institutional mandates. The point is not that the underlying abuses are invented—they are not—but that the conceptual packaging is itself a choice with consequences for how the phenomenon is measured and governed.

Two of the umbrella’s constituent forms sit awkwardly beneath it. Child labour counts as modern slavery only where it crosses into coercion or the worst forms identified in the ILO’s Convention No. 182 of 1999; ordinary, if undesirable, child work does not, and the boundary is contested. State-imposed forced labour—prison labour beyond lawful limits, abusive military conscription, and the kind of mass ‘labour transfer’ documented in Xinjiang—is estimated to affect several million people, yet it was deliberately excluded from the ILO’s 2024 study of forced-labour profits for want of reliable data,¹⁴ a telling admission that even the headline institutions know where their evidence runs thin.

Governments have layered their own vocabularies on top. The United States treats ‘trafficking in persons’, ‘human trafficking’ and ‘modern slavery’ as interchangeable umbrellas under its Trafficking Victims Protection Act of 2000, covering both sex trafficking and compelled labour.⁶ The United Kingdom and Australia went further, enacting statutes that use ‘modern slavery’ itself as the operative term. The advantage of the umbrella is breadth; its cost is analytical slippage. When a single word can denote a child soldier, a bonded brick-kiln worker, a trafficked sex worker and a woman married against her will, aggregate figures become harder to interpret and easier to weaponise. The Australian Anti-Slavery Commissioner, Chris Evans, draws the line at the point of exit—people are forced to work through coercion, deception or threats, but the defining feature is that they cannot leave:

“They cannot leave—that is the difference between modern slavery and severe exploitation.”⁷Chris Evans, Australian Anti-Slavery Commissioner

It is a usefully strict test, and one that much of the headline data does not consistently apply.

TWO The arithmetic of the unfree

The figure that anchors almost every discussion is 49.6 million: the number of people estimated to be in modern slavery on any given day in 2021, of whom 27.6 million were in forced labour and 22 million in forced marriage. It is the product of the Global Estimates of Modern Slavery, a joint exercise by the ILO, the human-rights group Walk Free and the International Organization for Migration.⁸ The estimate rests on nationally representative household surveys—sixty-eight covering forced labour and seventy-five covering forced marriage—combined with the IOM’s case data on identified trafficking victims.⁹ On its own terms it is the most ambitious attempt yet made to measure the immeasurable.

The same exercise reports an increase. The 2021 figure is roughly ten million higher than the 2016 estimate, an apparent worsening the authors attribute to a compounding of crises: the COVID-19 pandemic, armed conflict and climate disruption, which together drove people into poverty, unsafe migration and dependency.¹⁰ The distribution confounds the comfortable assumption that slavery is a problem of distant, poor states. More than half of all forced labour, and a quarter of all forced marriages, occur in upper-middle-income or high-income countries, and the overwhelming majority of forced labour—around 86 per cent—is imposed not by states but by private actors in the ordinary economy.¹¹ Women and girls are disproportionately represented, especially in forced marriage and in commercial sexual exploitation.

Yet it is here that the sceptic must slow down. The lead figure is not a count; it is a model. The anti-trafficking scholar Anne Gallagher, among the most authoritative critics of the field, has argued for years that prevalence estimates of this kind extrapolate from a limited set of country surveys to nations where no comparable data exist, that their authors have tended to overstate the reliability of the result, and that the appetite for a single arresting headline number has at times outrun methodological caution.¹² Her critique is not a denial that the phenomenon is large. It is a warning that precision is being claimed where only estimation is possible, and that a contested model, once it becomes a benchmark, tends to harden into received fact through sheer repetition.

The weaknesses are knowable even from the inside. Forced marriage is the hardest of the categories to quantify, because the line between a marriage that is coerced and one merely arranged within customary norms is drawn differently across cultures, and respondents may not classify their own experience as either. Forced-labour surveys, for their part, struggle to reach exactly the people most deeply trapped—those held incommunicado in compounds, detention or remote worksites—so the figure may be undercounted at its most severe end, even as broad screening questions risk sweeping in milder exploitation at the other. The published estimate carries no confidence interval prominent enough to discipline the way it is cited, and the comparison with 2016 is complicated by refinements in method between the two rounds. A number can be the best available and still be a poor foundation for the precise claims of growth routinely built upon it.

Two propositions can therefore be held at once. First, the existence of forced labour and forced marriage at a very large scale—tens of millions of people—is corroborated across independent bodies and is not seriously in dispute. Second, the specific figure of 49.6 million, and the confidence with which year-on-year ‘increases’ are asserted, deserve to be treated as informed estimates carrying real uncertainty rather than as a census. The growth the headlines describe is partly a change in the world and partly a change in our instruments for observing it.

THREE Why it persists

If the scale is real, its causes are depressingly intelligible. Exploitation flourishes where vulnerability concentrates, and the drivers of vulnerability are familiar. Poverty narrows choices to the point where a coercive job is preferable to none. Migration—particularly irregular migration, where the migrant has no legal standing and no recourse—hands leverage to recruiters and employers. Conflict shatters the institutions that might otherwise protect people, while displacement strips them of documents, networks and bargaining power.

The most recent data from the UN Office on Drugs and Crime sharpen the picture. Its 2024 Global Report on Trafficking in Persons, covering 156 countries, recorded a 25 per cent rise in detected victims in 2022 compared with pre-pandemic 2019, with trafficking for forced labour up 47 per cent and the number of detected child victims up 31 per cent.¹³ The agency links these movements explicitly to poverty, conflict and the climate crisis, noting in particular that in Africa the volume of climate-driven displacement correlates with the number of African trafficking victims detected in Europe. Weak governance and corruption supply the permissive environment; organised crime supplies the organisation; discrimination decides who is considered expendable. None of these factors is new. What has changed is their intensity and, crucially, the capacity of authorities to detect their consequences—which is why a rise in ‘detected victims’ is such a treacherous statistic, capable of reflecting either more crime or merely more competent counting, or both.

The mechanism that converts vulnerability into bondage is, more often than not, debt. Workers pay recruiters for the promise of a job abroad, arrive owing sums they cannot repay, and find their wages withheld, their documents confiscated and their movement curtailed until a debt engineered only ever to grow is settled. Labour-market precarity does the rest: where work is informal, subcontracted and lightly regulated, the distance between a bad job and a coercive one shrinks to almost nothing, and the gig and platform economies have in places widened rather than closed that grey zone. Vulnerability is not a metaphor; it is a balance sheet.

FOUR Following the money

Exploitation persists, ultimately, because it pays. In 2024 the ILO put the annual illegal profits generated by forced labour in the private economy at 236 billion US dollars, an increase of 64 billion, or 37 per cent, on its previous estimate a decade earlier.¹⁴ The report opens without euphemism:

“$236 billion. This is the obscene level of annual profit generated from forced labour in the world today.”¹⁵ILO, ‘Profits and Poverty’ (2024)

The composition of that sum is itself revealing. Forced commercial sexual exploitation accounts for roughly three-quarters of the profits while representing only about a quarter of private-sector victims, a disparity explained by the brutal economics of per-victim yield: the ILO estimates traffickers extract on average close to 10,000 dollars per victim across all forms, but far more from sexual exploitation than from labour exploitation.¹⁶ The remaining profit is harvested across the industries that recur, year after year, on every watch-list: agriculture and fishing, construction, mining, domestic work, manufacturing, fast fashion and hospitality. These are not exotic sectors. They are the sinews of ordinary consumption, which is exactly why the exploitation embedded in them is so difficult to excise—and why it has become a matter of trade policy rather than charity.

The renewable-energy supply chain is the sharpest contemporary illustration. Xinjiang produces a large share of the world’s solar-grade polysilicon—estimates cluster around 45 per cent—and Sheffield Hallam’s investigators found that all four polysilicon manufacturers in the region were implicated in coercive labour transfers, directly or through their raw materials, with some ninety Chinese and international companies exposed somewhere in their chains.¹⁷ The same logic of cheap, coerced inputs that once applied to cotton and cocoa now applies to the hardware of decarbonisation. Profit accumulates wherever the cost of labour can be driven below its market price by force, and the longer and more opaque the supply chain, the easier that compression is to hide.

Two features explain why these profits prove so durable. The first is debt bondage through recruitment fees, which lets an employer capture not merely a worker’s labour but the worker’s future, and which the ILO identifies as a principal engine of illegal gain. The second is structural: exploitation is not an aberration within these industries but, in places, a competitive equilibrium. A distant-water fishing operator whose crew works unpaid for months at sea, or a fashion brand whose third-tier subcontractor uses bonded labour, enjoys a cost advantage that scrupulous competitors cannot match without losing the contract. Eradication therefore asks firms not simply to behave better but to surrender an edge their rivals may still take—which is why voluntary virtue has so consistently disappointed, and why the argument keeps returning, however uncomfortably, to compulsion.

FIVE The digital mutation

If one form of modern slavery captures the novelty of the present moment, it is the cyber-scam compound. Across Myanmar, Cambodia and Laos, organised criminal groups—many originating in China and exploiting the lawlessness of Myanmar’s civil war—have built industrial fraud operations staffed by trafficked workers. The UN Human Rights Office reported in 2023 that credible sources indicated at least 120,000 people might be held in such conditions in Myanmar and around 100,000 in Cambodia.¹⁸ Victims, often educated and multilingual, are lured by fraudulent job advertisements, confined behind razor wire, and forced to run romance and cryptocurrency-investment scams against strangers on other continents. Those who fail to meet targets are, by repeated account, beaten or tortured.

The High Commissioner for Human Rights, Volker Türk, framed the moral complexity precisely: the people coerced into committing online fraud are not perpetrators but captives.

“They are victims. They are not criminals.”¹⁹Volker Türk, UN High Commissioner for Human Rights

The scale is contested but plainly vast. The UNODC has estimated that Southeast Asian scam centres generate close to 40 billion dollars a year; the United States Institute of Peace suggested that by 2024 around half a million people were forcibly involved, with the proceeds amounting to a striking share of the formal economies of the host states.²⁰ Amnesty International, after interviewing more than 400 survivors, documented abuses on a ‘mass scale’ across at least fifty-three Cambodian compounds, including the use of electric shocks.²¹ The phenomenon fuses two crimes at once—the trafficking of the workers and the defrauding of their victims—and it is here that technology has most obviously transformed exploitation. Recruitment runs through social media; coercion is enforced inside guarded compounds; and the loot moves as cryptocurrency, which is precisely why the United States has reached for financial sanctions, with its Treasury designating networks tied to compounds after Americans lost an estimated 10 billion dollars to such scams in a single year.²² Artificial intelligence cuts both ways: it lets traffickers automate persuasion and translation at scale, even as investigators deploy it to map laundering networks and identify victims. The technology is neutral; the advantage, for now, is contested.

What makes the scam compound so resistant is the doubling of harm it embodies. The trafficked worker and the defrauded mark are victims of the same syndicate, yet they are set against one another, and enforcement systems built to pursue fraudsters struggle to grasp that the ‘fraudster’ at the other end of the message is themselves a captive. So-called ‘pig-butchering’ schemes—in which a target is fattened with affection and fictitious investment returns before the slaughter—depend on patient, sustained human labour, which is precisely why the syndicates require a coerced workforce. Dismantling the model therefore means rescuing the workers as surely as protecting the targets, a duality that fits poorly with the instincts of either fraud squads or border control.

SIX Trade as a weapon, and the Australian test case

The shift that gives this subject its current political charge is the conversion of forced-labour concern into trade enforcement. The legal hinge in the United States is old: Section 307 of the Tariff Act of 1930 bars the import of goods made wholly or in part by forced labour. For decades it was barely used. The Uyghur Forced Labor Prevention Act of 2021 reanimated it by creating a rebuttable presumption: any goods originating in Xinjiang are presumed to be made with forced labour and barred from entry unless the importer can prove otherwise by clear and convincing evidence.²³ By August 2025 US Customs and Border Protection reported having stopped more than 16,700 shipments worth almost 3.7 billion dollars for examination under the Act, denying entry to over 10,000 of them, with 144 companies on its Entity List.²⁴ The 2025 enforcement strategy, issued under an administration that frames the policy in frankly competitive terms—“America First means keeping foreign goods made with forced labor off our shelves”, in the words of the Secretary of Labor²³—widened the net to new high-priority sectors including steel, copper, lithium and caustic soda.

This is where the sceptic earns their keep. The documentary basis for forced labour in specific Xinjiang supply chains is among the strongest evidence in the entire field; it is granular, corroborated and traceable. Yet the same instrument serves a geopolitical and protectionist agenda, and its enforcement has been erratic. Analysts at the Center for Strategic and International Studies noted that detentions fell sharply in the first half of 2025, with only fourteen shipments stopped in July and no additions to the Entity List for months—raising the awkward possibility that the rhetoric of moral seriousness and the reality of administrative effort had diverged.²⁴ A concept that is simultaneously a human-rights standard and a tariff lever will always invite the suspicion that the standard is being applied selectively, against rivals rather than wherever the evidence leads.

Australia offers a cleaner test of intentions, and a sobering one. In 2018 it became the second country, after the United Kingdom, to legislate against modern slavery, requiring large entities to publish annual statements describing the risks in their operations and supply chains.²⁵ The model was transparency rather than enforcement—and that was its weakness. The statutory review led by Professor John McMillan, tabled in 2023, made thirty recommendations and reached a blunt conclusion: the reporting-only approach had raised awareness but had not reliably translated into accountability, and there was no hard evidence that the Act had yet changed the lives of those actually in conditions of modern slavery.²⁶ McMillan urged penalties for non-compliance, a lower reporting threshold and a genuine due-diligence obligation. The government accepted most of the recommendations in principle in December 2024 and appointed the country’s first federal Anti-Slavery Commissioner, Chris Evans, though the Act still carries no penalties for the companies it regulates.²⁷ Walk Free’s modelling places roughly 41,000 people in conditions of modern slavery within Australia itself—forced marriage among the most frequently reported forms—even as Australia is ranked among the small group of ‘Tier One’ states in the US Trafficking in Persons report.²⁸ The lesson is instructive for every jurisdiction tempted by disclosure laws: a statute that asks companies to describe a problem, without requiring them to fix it or punishing them for ignoring it, produces excellent paperwork and uncertain results.

Set against its peers, Australia’s instrument looks characteristic of the transparency generation, and increasingly dated beside what is coming. The United Kingdom moved first, in 2015, with a Modern Slavery Act whose supply-chain transparency provision Australia closely echoed—and which has drawn the same criticism for relying on disclosure without sanction, even as Britain installed an Independent Anti-Slavery Commissioner to press the system along.²⁹ The European Union has chosen a more muscular path: a Corporate Sustainability Due Diligence Directive imposing binding obligations to identify and address harms, paired with a regulation barring forced-labour goods from the single market outright. The United States, distinctively, runs two engines at once—criminal prosecution under its trafficking statute and the border enforcement of Section 307—which is why its measures bite hardest on trade. Australia, with disclosure but as yet no penalties, sits conspicuously at the softer end of that spectrum.

Within Australia, the texture of the problem is its own, and it confounds the assumption that slavery is something that happens elsewhere. Forced marriage is consistently among the most frequently reported modern-slavery matters to the Australian Federal Police; in 2024–25 the force received around 420 reports of human trafficking and modern slavery in total, nearly double the figure of five years earlier, with forced marriage and ‘exit trafficking’—removing a person from the country for exploitation—accounting for the largest increases.³⁰ Strikingly, reports of forced labour remained low and even fell, despite well-documented concerns in sectors such as horticulture, where migrant and seasonal workers on visas that tie their lawful presence to a single employer are structurally exposed, the threat of deportation operating as its own form of coercion. The Anti-Slavery Commissioner himself has described that gap between suspected prevalence and actual reporting as deeply concerning. The vulnerabilities, in short, are domestic as well as imported, and a statute trained on the overseas supply chains of large corporations reaches only part of them.

SEVEN Voices from the front line

The institutions that produce the figures are not naive about their meaning, and their public statements repay close reading. When the 2021 estimates were released, Walk Free’s founding director, Grace Forrest, described modern slavery as the antithesis of sustainable development and ‘a man-made problem, connected to both historical slavery and persisting structural inequality’—a framing that locates the cause in economics rather than in individual villainy.¹²

The IOM’s then director-general, António Vitorino, used the same occasion to press a narrower, evidence-based point: that the vulnerability of migrants to forced labour depends above all on whether national legal frameworks protect people regardless of migration status, and that safe and regular migration is therefore a frontline defence rather than a soft afterthought.¹² These are not the cadences of activism. They are the carefully bounded claims of officials who know their numbers will be challenged, and who confine themselves to what the data will bear. Set beside Anne Gallagher’s methodological scepticism and the survivor testimony gathered by Amnesty and the UN Human Rights Office, they compose a field that is, at its best, considerably more self-critical than its detractors allow—and considerably more honest about uncertainty than the headlines it generates.

Officials closer to the casework are blunter about the distance between recognition and response. Presenting the UNODC’s 2024 findings, its representatives stressed both the sharp rise in child trafficking and the inadequacy of the criminal-justice systems meant to counter it—a reminder that detection without prosecution is only half a remedy.¹³ The survivor testimony assembled by the UN Human Rights Office and by Amnesty International supplies what the aggregates cannot: the dark rooms, the confiscated passports, the debts designed to be unpayable, the recruiter who was once a friend. It is in the gap between the statistician’s careful estimate and the survivor’s specific account that the honest reader must calibrate how much of this crisis is genuinely new and how much is merely, at long last, being seen.

EIGHT Are we winning?

The honest answer is: not yet, and not obviously. There are real gains to record. Victim identification has improved, which is part of why detection figures are rising. Supply-chain transparency laws have spread from Britain and Australia to the European Union, whose forthcoming due-diligence regime will impose obligations of a different order. Corporate behaviour has demonstrably shifted in response to US enforcement; companies have re-engineered supply chains, sometimes within a single year, to escape the reach of the Uyghur Forced Labor Prevention Act—evidence that credible penalties move markets in a way that voluntary disclosure does not.²⁴

Against this stands the weight of the countervailing evidence. The aggregate estimates have risen, not fallen. Illegal profits have grown by more than a third in a decade. An entire new modality of forced labour—the scam compound—has materialised and scaled within a few years. Prosecutions remain rare relative to the scale of the alleged crimes; the ILO has repeatedly noted how few jurisdictions secure convictions for forced labour. And the disclosure statutes that looked, a decade ago, like a decisive intervention have produced, in the candid words of Australia’s own statutory reviewer, no hard evidence of changed lives. Awareness has risen faster than outcomes.

The prosecution gap is the starkest measure of that mismatch. Detection, identification and disclosure have all advanced; conviction has not kept pace. Australia supplies a chastening illustration: it criminalised forced marriage in 2013, yet by the end of 2023 not a single such case had yielded a successful prosecution and conviction, even as reports to police climbed year upon year.³¹ Multiply that attrition across jurisdictions and the picture is of systems increasingly able to name the crime and persistently unable to punish it. Where impunity is the norm, the deterrent on which every legislative scheme implicitly relies quietly evaporates, and disclosure obligations become a substitute for justice rather than a route to it.

NINE What would actually work

The reform debate has, encouragingly, begun to converge on measures with an evidentiary basis rather than a rhetorical one. The clearest pattern is that obligations bite where they carry consequences. Mandatory human-rights due diligence—requiring companies not merely to report risks but to identify, prevent and remediate them, on pain of liability—consistently outperforms voluntary transparency. Import bans on forced-labour goods, for all their selective application, have demonstrably altered corporate sourcing in ways that two decades of disclosure did not.

Beyond the supply chain, the interventions best supported by evidence are unglamorous. Properly resourced labour inspection reaches the informal sectors where exploitation concentrates. Worker-driven models, in which the people most exposed to abuse design and monitor the standards that protect them, have shown durable results in agriculture. Migration reform that gives workers legal status and the freedom to change employer removes the single greatest source of leverage traffickers possess. Survivor-led interventions improve both identification and the quality of remedy. Technology-assisted detection—mapping financial flows, scraping recruitment networks—can multiply the reach of overstretched investigators. None of these is a panacea, and the field’s besetting temptation is to mistake the measurement of a problem for its solution. The metric is not the remedy.

END Seeing what was always there

So: is modern slavery genuinely increasing, or are we finally measuring an old and constant evil? The most defensible answer is that both are true, in proportions we cannot yet cleanly separate. The structural drivers—poverty, conflict, displacement, climate disruption, irregular migration—have intensified in the twenty-first century, and it would be implausible to suppose that exploitation had not grown with them. The cyber-scam compound is a genuinely new phenomenon, not a measurement artefact. To that extent, the rise is real.

But a large share of the apparent increase is the product of better instruments: expanded survey coverage, new legal categories, improved victim identification and a public vocabulary—‘modern slavery’—that did not exist a generation ago and now draws attention to practices long tolerated in plain sight. The headline figure of 49.6 million should be carried lightly, as the contested estimate it is, while the underlying reality it gestures toward—exploitation at a scale of tens of millions, woven into the legitimate economy—should be taken with complete seriousness. The strongest evidence in the entire field is not the global aggregate at all but the granular, traceable documentation of specific supply chains, which is why trade enforcement, for all its geopolitical entanglements, rests on firmer ground than the round numbers do.

That firmer ground carries its own hazard. To make forced labour a criterion of trade is to fuse a humanitarian standard with an instrument of economic statecraft, and the two do not always pull in the same direction. The same border that excludes a tainted solar panel can be used to disadvantage a strategic rival under cover of conscience, and enforcement that surges and subsides with the diplomatic weather invites exactly that suspicion. A standard applied only against adversaries is not a standard but a tactic. The credibility of the whole enterprise depends on whether the principle is allowed to follow the evidence wherever it leads—including into supply chains closer to home, and into the seasonal fields and suburban homes where exploitation is least convenient to find.

The question that opened this essay—is slavery returning, or are we finally seeing it?—may therefore be the wrong dichotomy. The more exact formulation is that a constant, very large substrate of human exploitation is being simultaneously enlarged by the pressures of the age and illuminated by the tools of measurement and law. We are seeing more because there is more to see, and because we have, at last, switched on the light. What remains conspicuously underdeveloped is not our capacity to count the unfree, but our willingness to bear the economic cost of setting them free.