Reading the İHD 2025 Report Through a Class Lens

The Other Side of the Balance Sheet: Reading the İHD 2025 Report Through a Class Lens

The Human Rights Association’s (İHD) 2025 rights-violations report counts 2,749 deaths. Of these, 2,105 died at work. The report gives us violations under rights headings; we will rearrange the same numbers under class headings. A very different picture emerges.

Dear Young Comrades,

At a press conference in Istanbul on 4 September 2026, the Human Rights Association released its 2025 Rights Violations Report. A 218-page main text and an 11-page balance sheet. As every year, the report was summarized in the press with a few striking figures and dropped from the agenda: 2,749 deaths, 8,477 detentions, 402,000 prisoners, 42 imprisoned journalists.

In this essay we want to catch the report before it drops from the agenda and stay with it a little longer. Because these figures, compiled by İHD’s labor, read with the right question, are the 2025 balance sheet of class struggle in Turkey. Not merely “rights,” but a record of whose right was violated, by whom, and in what relation.

Let us say this clearly at the outset: the work İHD has done is valuable, and this essay is built on it. Without the report, this essay would not exist. What we want to do is make visible what the report’s own categories do not show us: the class distribution of the data.

We will do three things. First we will discuss the report’s own frame and the limit of that frame. Then we will rearrange the figures through a class lens. Finally we will look at what tasks this table lays on us.

Part One: How the Report Counts, How We Will Read

Rights categories and what they do not say

The İHD report is organized by the categories of international human-rights law: the right to life, the prohibition of torture, personal liberty, freedom of assembly and demonstration, freedom of expression, organization, participation in government, the prohibition of discrimination, refugee rights, prisoners’ rights, the right to the environment, impunity. This is a universal and comparable language; it makes it possible to set a Turkey report beside a Hungary report. The strength of this language is there.

But this language also has a structural silence. Human-rights categories define the violation as a relation between the individual and the state. The subject is the abstract “human,” the addressee the abstract “state.” In this frame, the worker who dies at work and the citizen killed at a demonstration are written into the same column, the “right-to-life violation” column. The social relation in which the death took place — who profited, whose wage was pressed down, who failed to inspect — does not appear in the category.

Marx made a basic observation on this as early as 1843 in On the Jewish Question: the “rights of man” proclaimed by bourgeois revolutions are the legal form of an order in which the citizen declared abstractly equal in the political community remains concretely unequal in civil society. This does not mean rights are worthless; Marx himself hailed factory laws and the limitation of working hours as “the victory of the political economy of the working class.” It means this: rights are gains of class struggle, and when read independently of the class relation their content empties out.

So our question will differ a little from the report’s. The report asks “which right was violated.” We will ask “in which social relation did this violation take place, and where did each party stand in that relation.”

Method: the same numbers, different columns

To do this we will not add new data to the report. We will take the report’s own figures and regroup them. A class reading is not an interpretation from outside the data; it is naming an order already inside the data.

Part Two: The Class Distribution of the Figures

1. The largest right-to-life violation is the workplace

According to the report, 2,749 people died in 2025 under the heading of right-to-life violations. Let us open this number by the report’s own subheadings:

Workplace murders: 2,105
Femicides: 299 women, 64 children
Deaths from state neglect and lack of oversight (fires, etc.): 159
Military operations and clashes: 70 (68 in the Kurdistan Region of Iraq)
Suspicious deaths (prison, detention, military service): 31
Political, racist, and phobic attacks: 15

One look at this list is enough. Three out of every four deaths (77 percent) took place at work. The number of workers killed in workplace murders is thirty times the number killed in clashes. At least five or six workers a day.

With İSİG Assembly data, the report records that this is the highest figure since the pandemic, that deaths in workplace murders between 2016 and 2025 exceeded 20,000 (20,009), and that since 2013, 24,860 workers have died while working. In twelve years, as many people as a town’s population went to work and did not come home.

Engels gave these deaths a name in 1845 in The Condition of the Working Class in England: social murder. He said: society knowingly places hundreds of proletarians in conditions that would count as murder if done with a knife or a bullet; if it knows death is approaching and still maintains the conditions, that is murder — yet because no one sees the perpetrator and death looks natural, it is a secret murder. Exactly this thought stands behind İSİG Assembly’s and İHD’s saying “workplace murder” instead of “workplace accident.” The report also writes it plainly: in a significant share of workplace murders, “the state’s failure to fulfill its duty of oversight is the greatest factor.” The cosmetics workshop in Dilovası where seven workers died on 8 November was illegal and unlicensed; the workers were uninsured.

Here the state is absent. Let us keep that “absence” in mind; in a moment we will look at where the state is more than present.

2. Looking at the ages of the dead makes class structure visible

The two ends of the workplace-murder data tell what is happening in the reproduction of labor-power in Turkey.

At one end, children: in 2025 at least 94 children died in workplace murders, 26 of them 14 and under. The report notes that this is the highest figure İSİG has recorded since 2011 and names the cause: the marketization of education, shortening school time through open high school, promoting vocational high schools under the name of “industry–education collaboration,” and putting children to work in enterprises at least four days a week through the Vocational Training Center (MESEM) model. In the report’s words, these are “different stages of again placing child labor massively at the service of capital.” That we are reading this line in a human-rights report shows how blatant the reality has become.

At the other end, the elderly: 684 workers aged 50 and over, 126 aged 65 and over, died while working. The report ties this to the fall in pensions and adds: one in every three workers who died was someone over 50 who either had not won the right to retire or, having won it, still had to work because they could not live on it.

Read these two ends together. Capital is pushing labor-power outside its “normal” age range in both directions at once: downward, to those who are still children; upward, to those who should long since have retired. When Marx in Capital describes the limits of the production of absolute surplus value, he describes exactly these two movements: lengthening the working day and expanding the working population to include women and children. Turkey’s 2025 labor market is reproducing the landscape in nineteenth-century English factory inspectors’ reports.

And a third group: 91 refugee and migrant workers. The report notes that Syrian refugees are “employed cheaply and without security under conditions exempt from every occupational health and safety measure.” When we add the 48 people who died trying to cross the border in the Aegean, we see that three-quarters of refugee deaths are also workplace murders. In Turkey the migrant is first of all not a “migration problem” but the most unprotected, cheapest layer of the reserve army of labor; the death statistics confirm it.

3. Wages: the largest violation, the uncounted violation

The report’s introduction gives the economic picture in short, precise sentences: broad unemployment in December 2025 at 28.6 percent, the number of unemployed past 11.5 million. Of the minimum wage left unraised through the year, 30.8 percent was eaten by inflation and deductions; wages in general fell by an average of 53 percent. In January 2025 the minimum wage was below the hunger line.

These figures are not counted as a “violation” in the report’s balance-sheet table, because the method of human-rights law does not allow it. Half a wage evaporating in a year is not something a particular perpetrator did to a particular person on a particular date; it is the continuous operation of a social relation. But through a class lens, 2025’s most massive rights violation is this: tens of millions of wage-earners paid below the value of their labor-power.

And this violation is the ground of the others. When wages fall, overtime lengthens; when overtime lengthens, workplace murder rises. As pensions erode, the older worker dies. When household income collapses, the child leaves school for the enterprise. When the report defines workplace murders as “the heaviest result of working conditions worsened by the economic crisis and of lengthening working hours that accompany the pressing down of wages,” it is already building this causality. We are only underlining it.

4. Two faces of the state: absent in the factory, present in the square

Now let us return to the clue we left above. The state was absent as oversight in the workplace where the worker died. So where was it present?

The report gives the answer. In 2025:

268 demonstrations were intervened against; in the interventions 8,477 people were detained, 131 injured.
With 19 general ban decisions, demonstrations were banned for a total of 126 days in places inhabited by 40 million 834 thousand people. Nearly half of Turkey’s population.
Of the 1,842 people subjected to torture and ill-treatment, 1,493, or 81 percent, experienced it during police intervention against a demonstration. In the report’s words, “torture and ill-treatment spilled into the street.”
2 strikes were postponed for 60 days by presidential decision as “of a nature to disrupt general health and national security,” that is, banned in practice. One was Öz Büro-İş’s strike at TSE, one the strike of the Turkey Mine Workers’ Union.
For union activity 28 people were detained, 3 arrested. BİRTEK-SEN General President Mehmet Türkmen was arrested while meeting textile workers who had walked out in Antep and held under house arrest for months. Eğitim-Sen officials were put under house arrest for calling for a boycott of classes. The joint headquarters of Enerji-Sen and Basın-İş was raided by police.

The table is this: the same state is missing as “lack of oversight” where capital kills the worker, and more than present as strike bans, detention, and tear gas where the worker organizes against capital. This is not an inconsistency; it is a consistency. The class character of the state appears not as a conspiracy theory but precisely in this asymmetry — which relation it intervenes in and which it does not.

The mining question compresses this into a single year’s example. In 2025 the mine workers’ strike was banned on grounds of “national security.” The same year, in the report’s words, “every legal and bureaucratic obstacle limiting mining activity” was removed by Law No. 7554; EIA processes were narrowed at companies’ request; urgent expropriation was applied in Akbelen; final authority over mining permits was transferred to a board chaired by the Vice President. The report also writes the reason plainly: “to accelerate the search for new sources by capital groups that rose in the AKP years.” So the worker cannot strike in the mine, but the company can enter the forest. The class content of “national security” is defined by setting these two decisions side by side.

5. Prison: a class indicator of four hundred thousand people

The report’s prisoner section is the section that most needs a class reading. As of 31 December 2025 there are 402,014 prisoners in Turkish prisons; 63,905 of them are remanded. In 2013 the figure was 144,212. In twelve years, nearly threefold. In the same period prison capacity rose 50 percent while the prisoner count rose 200 percent; occupancy 131.7 percent, 96,728 people over capacity. 4,421 child prisoners. At least 1,412 sick prisoners, 335 of them seriously.

We are used to looking at this number through the lens of “political prisoners”; rightly, because political prisoners are the most visible face of the prison question. But the great majority of the 402,000 are not political but ordinary criminal prisoners. And one need not have studied criminology to know from which class the ordinary prison population is drawn: theft, drugs, fraud, checks and promissory notes, alimony. The translation of poverty into crime categories.

Historically the prison has functioned as a depot of the reserve army of labor. The “bloody laws” Marx recounts in the primitive-accumulation chapter of Capital punished peasants torn from the land as vagrants, beggars, and thieves. In Turkey, 2013–2025 is the period of falling wages, spreading precariousness, postponed retirement rights, and broad unemployment nearing 11 million. That the prison population tripled in the same period suggests these two curves are not independent. The report does not draw this link explicitly, and could not be expected to; but a class reading must ask exactly this link.

And the prison is working too: prisoners at 32 percent over capacity, 58 people in a 28-person ward (Sincan), 68 in another example. People held in these conditions are kept waiting to return to the lowest rung of the labor market outside. A criminal record is the most reliable ticket to precarious work.

6. The political field: a fight within capital, or a war on the class?

2025’s most visible political event was the cancellation of Istanbul Metropolitan Mayor Ekrem İmamoğlu’s diploma, his detention, and his arrest. According to the report, in the nine days after the detention 1,879 people were detained, 260 arrested. Through the year 27 mayors were removed from office, 16 arrested, trustees appointed to 13 municipalities. The number of voters represented by the arrested mayors was 14 million 870 thousand, more than a quarter of all voters. 1,527 CHP members and officials suffered rights violations; 316 people from the DEM Party faced judicial harassment; 222 from ESP were detained. Show TV, Habertürk, BloombergHT, Flash TV, Tele1, and Gain Medya were transferred to the TMSF.

How will we read this? The liberal reading calls it a “slide from democracy to authoritarianism”; that is a true but incomplete description. A class reading must distinguish two layers.

The first layer is a distributional fight inside the ruling bloc. Municipalities are among the most important channels of capital accumulation in Turkey: zoning, tenders, land, transport, waste, advertising. The changing hands of municipalities and the redistribution of media ownership through the TMSF are a fight over who will accumulate capital. In this fight the working class is not a party; it is a spectator or material.

The second layer is the form in which this fight is conducted, and here the working class is a direct target. Because even a fight inside the ruling bloc, when conducted by closing streets, banning demonstrations, arresting unionists, and turning speech into crime, closes a field that belongs to everyone. A demonstration ban applied to 40 million people may have been issued for İmamoğlu; but under that ban there is also the textile worker who cannot strike. In Gramsci’s language: as economic crisis erodes consent, rule rests more and more on force; once the coercive apparatus expands, only the power of the day’s need decides against whom it will be used. Today against the mayor, tomorrow against the strike committee.

So it is a mistake to see 2025’s political crisis as an “extra-class” matter. But putting it in place of class struggle under the heading “democracy struggle” is an equal mistake. The right stance is to struggle against the narrowing of the political field while asking for whom and with what that field will be filled.

7. Peace and repression in the same year

There is one more contradiction the report underlines. 2025 was the year hope for peace rose most on the Kurdish question: Öcalan’s call on 25 February, the PKK’s dissolution decision on 12 May, the disarmament ceremony on 11 July, a commission in Parliament. The report records these in order. And it immediately adds: “the failure to take democratic steps that would nourish and strengthen the climate of peace, and moreover the continuation of a governing outlook that makes it a principle to crush every social reaction with the sword of the judiciary and the baton of the police, again led to the violation of many fundamental rights.” Trustees were appointed to eight DEM Party municipalities; judicial harassment of DEM Party members continued; a security zone was declared in Şırnak for 175 days.

We wrote in the Peace Dossier: capital’s peace is the silencing of weapons; labor’s peace is the presence of justice. This report is the proof of that distinction with 2025 data. In the year weapons fell silent, demonstration bans, strike postponements, and trustees did not decrease. This shows that the class content of the peace process has not yet been filled from below; peace from above does not feel obliged to lift repression below.

8. Impunity: law as class justice

The report’s final section is devoted to impunity, and this section is like the keystone of the whole essay. The upholding on appeal of the acquittal of police in the Tahir Elçi case; a fresh acquittal despite the Constitutional Court’s retrial decision in the case of Nezir Tekçi, disappeared in detention in 1995; the dropping for “prescription” of the case against 196 gendarmes over the deaths of 12 people at Bayrampaşa in the 2000 “Return to Life” operation. The report calls impunity “an unnamed state policy.”

Let us also look at this from the other direction. Under insulting the president (TPC 299), cases were opened against 234 people, 194 detained, 65 arrested. According to Ministry of Justice data, under TPC 299–300–301 cases were opened against 8,005 people, 9,348 cases filed, 12,414 cases ongoing. 298 detentions for social-media posts. For a citizen who criticizes the government in a street interview, the law is fast, definite, and hard; for an officer who kills a person in detention, the law is slow, indefinite, and inclined to prescription.

The name of these two speeds together is class justice. To say that law is the will of the ruling class clothed in the form of statute does not mean the law always takes sides in a crude way; it means the same law operates at different speeds for different subjects. The report has measured this; we are only saying what the measurement means.

Part Three: Two Readings, One Report

Let us set what we have said so far side by side. The left column is the report’s own frame; the right column is the class re-reading of the same data.

	Rights-based reading	Class reading
Subject	Abstract individual / citizen	Human in a concrete class position: worker, unemployed, refugee, prisoner
Addressee	The state	The capital relation and the state that protects it
Unit of violation	Event: particular perpetrator, particular victim, particular date	Relation: a continuously operating regime of exploitation and domination
2025’s largest violation	Demonstration bans, arrests, trustees (by order of visibility)	53 percent wage erosion and 2,105 workplace murders (by order of mass)
Place of workplace murder	A subheading under the right to life	The center of the whole table: 77 percent of deaths
Role of the state	Perpetrator of the violation or party obliged to prevent it	Absent in the factory, present in the square: selective intervention
Prison	Prisoners’ rights, conditions, sick prisoners	Depot of the reserve army of labor; translation of poverty into crime
Political crisis	Slide from democracy to authoritarianism	Distributional fight inside the ruling bloc + general expansion of the coercive apparatus
Impunity	Deficit of the rule of law	Class justice: different speeds of the same law
Horizon of solution	Return to human-rights standards	A class struggle that includes the standards but also surpasses them

This table is not to invalidate the left column but to show that the two columns need each other. The right column cannot speak without the left column’s data; the left column cannot find a cause without the right column’s question.

Part Four: What Does This Table Tell Us?

The question of scale

The first result of a class reading is a correction of scale. The images the public will remember from 2025 will be the crowds at Saraçhane, the water cannons, the arrested mayors. These matter, and this essay does not belittle them. But the report’s figures show that 2025’s real mass violence was lived where there is no image: in the workshop, on the construction site, in the mine, in the field, on the illegal floor of a cosmetics depot. And the perpetrator of that violence is not an identifiable police officer but a production relation left without oversight.

The question of causality

The second result is that the violations are linked. Wage erosion, lengthening overtime, workplace murder, child labor, elderly labor, refugee labor, prison population: these are not separate “violations” but different appearances of a single process. That process is the bill for the economic crisis being charged to labor. Strike bans and demonstration bans are the tools that ensure the bill is paid without objection. Impunity is the mechanism that keeps those who issue the bill safe.

The question of hope

The third result is written in the report’s own introduction: “Fortunately, there are those who object to all this dark course.” 8,477 detentions, 268 demonstrations, 1,493 people subjected to ill-treatment in the street are measures not only of repression but of resistance. No one bans what no one would do. If a demonstration ban must be placed on 40 million people, those 40 million have a chance of taking to the street.

Concrete Tasks

Use the report as a working text in your union, chamber, or association. You can download the full text and the balance sheet from İHD’s site. You need not read 218 pages in one sitting; the workplace-murder, union-rights, and prisoners’-rights sections are of a size that can be presented at a workplace meeting. Figures that feel distant when abstract come near when brought down to your own sector.
Take workplace murders out of the “right to life” heading and place them at the center of the labor agenda. After every workplace murder, ask not only about “accident” but about hours, wages, insurance, and oversight. Follow İSİG Assembly’s monthly reports; classify deaths in your region by sector, age, and status.
Join the fight against child labor to the fight over education. The MESEM model is a labor-market policy in the guise of an education policy. The link between education unions and workers’ unions is the first concrete door of the polytechnic education program we discussed in the Democratic Education Dossier.
Treat strike bans and demonstration bans not as separate matters but as two ends of the same mechanism. The “national security” that bans the mine worker’s strike and the “public order” that closes Saraçhane come from the same pen. The bridge between democracy struggle and class struggle is built on this common ground.
Speak of the prison question not only through political prisoners but through the whole of the 402,000. Demand data on prisoners’ class composition; Ministry of Justice statistics give the type of crime but not occupation and income. The gap itself is a political choice.
Organize the refugee worker not as a “migration problem” but as part of the class. 91 refugee worker deaths show racism’s function in the labor market: leaving the cheapest labor the least protected. Let us repeat what we said in the Racism and Class essay: a divided class is a class that loses the wage bargain.
Add your own balance sheet to the balance sheet. İHD’s data source is largely the press and applications; violations lived at work and not reflected in the press fall outside these figures. Unions and professional organizations keeping their own violation records and forwarding them to İHD will make next year’s report more class-based still.

Dear Young Comrades,

The İHD report counts 2,749 deaths in 2025. Of that number, 2,105 are people who left home for work in the morning and did not return in the evening. No one declared a curfew for them; no one’s diploma was cancelled; no case was opened against them. They died quietly and were counted quietly.

To read through a class lens is first to break that silence. Then it is to see that this silence is the product of the same order as the noisy repression in the street. The state that does not inspect in the factory and the state that fires tear gas in the square are the same state; the decision that bans the strike and the law that opens the forest to the mine are the same will; the law that frees the torturer by prescription and the law that arrests for a social-media post are the same law.

The report gives us all of this. Naming it is left to us.

And naming is not enough. For each of the 2,105 workers who died there is a workplace, a sector, a question of organization. The answers to those questions will determine what the number will be in next year’s report.

A balance sheet is not an accounting document; it is a call.