The Week's Class Struggle in Technology

6–12 October 2026

October 12, 2026 at 11:14 AM43 min readAuthor: Oğuz Demirkapı

Series:The Week's Class Struggle in Technology(Part 1 of 1)

Class Struggle in Technology: The Important Events of the Week

The Week's Class Developments in Technology — Issue 1 · 6–12 October 2026

About this series. With this piece we are starting a new series. Every Monday we will gather the developments of the previous week in the world of technology that matter from a class point of view: strikes, union drives, layoffs justified by artificial intelligence, local resistance to data centres, algorithmic surveillance, digital censorship. Under each heading we will try to answer three questions: What happened? What does it mean in class terms? Why does it matter for us?

We want to see not the news itself, but the relation underneath the news: who produces, who decides, who gains, who pays the bill. The headings link to the primary source of the news; inside the text there are links to the relevant concepts and to our earlier pieces under our artificial-intelligence tag.

First, a Short Summary

Dear young comrades,

Last week twelve developments took place in the world of technology that look independent of one another. But all of them turn around the same question:

In whose hands will the productivity, the wealth and the power created by artificial intelligence be concentrated; who will pay the cost of the transformation?

The week's table, in short, is this:

In Taiwan, Micron workers, who produce the memory chips of artificial-intelligence servers, won strike authorisation by 99 per cent. The demand: a permanent share of the profit.
In Norway, the largest bank, DNB, is removing about 400 posts, citing artificial-intelligence agents; "code writing" is on the list too.
In the United States, Amazon carried out a new round of white-collar layoffs; KQED examined why technology workers are turning to the union; in California, tens of thousands of platform drivers are meeting in a union; Wizards of the Coast workers want a binding rule against generative artificial intelligence.
In Australia, the engineers and technology specialists of the institution that runs the electricity market are extending their action to operational work.
In India, Amazon's and Google's giant data centres are meeting popular objection over water, electricity and consent.
In the United States, the surveillance-camera company Flock Safety is laying off a fifth of its workers as the social backlash grows; the Trump administration, by suspending technology companies' green-card applications for foreign workers, is setting worker against worker.
In Turkey, the TMMOB published its artificial-intelligence report; the wave of digital censorship reached EngelliWeb itself; at the opening of the academic year Erdoğan said that artificial intelligence "has put labour out of use".

Below we take up each of them in detail. Then we draw out the four axes that bind them together, and at the end we list the concrete tasks that fall to us.

The Week's Table: The Language of Capital, the Question of Labour

Development	The language of capital	The question of labour
Micron strike authorisation	"Record profit, artificial-intelligence demand"	If we are the ones who produce the profit, why is our share not permanent?
DNB, 400 posts	"Efficiency, the new reality"	Why is the return on a rise in efficiency unemployment?
Amazon layoffs	"Restructuring according to priorities"	Why is a period of record investment a period of layoffs?
Organising in Silicon Valley	"Stock option, talent, flexibility"	Does a good salary save you from being a worker?
The California drivers' union	"Independent contractor"	Who will object to the algorithm's decision?
Wizards of the Coast	"Tools of creative efficiency"	Whose is the product of creative labour?
AEMO actions	"Cost to the consumer, public safety"	Where will the labour that keeps critical infrastructure standing use its say?
Data centres in India	"Digital development, investment"	Who will drink the cloud's water?
Flock Safety	"Public safety"	Who pays the cost of surveillance?
Green-card suspension	"Protecting the American worker"	Protecting the worker, or dividing the worker?
The TMMOB report	"The national artificial-intelligence drive"	Technology to whose benefit, under whose control?
The EngelliWeb block	"National security, public order"	What is the sign of silencing the one who counts the censorship too?

I. On the Production Line: The Sharing of Profit

1. Strike Authorisation from Micron Workers: The Labour behind the Artificial-Intelligence Chips

7 October · Taiwan · Semiconductors

What happened? The union organised at the Taoyuan plants, in the north of Taiwan, of the US memory-chip maker Micron won the approval of 1,994 members, that is 99 per cent of those who voted, in a strike-authorisation ballot. The two unions in Taoyuan and Taichung together represent more than 80 per cent of Micron's roughly 15,000 workers in Taiwan. Mediation with the Taoyuan union ended in September without an agreement. According to Reuters, the union's basic demand is a permanent profit-sharing mechanism. The rewards the company announced for the 2026 financial year, including a cash bonus of 1 million Taiwan dollars (about 31,000 US dollars) for workers in Taiwan, do not meet this demand. The union also criticised the company's announcing these rewards unilaterally while mediation was still going on and before an agreement had been reached. A strike date has not yet been set; the union called on members to take leave in order to join the rally in Taipei on 19 October. Taiwan's Ministry of Labour said it respected the ballot and called on the company to bring concrete proposals. When the news broke, Micron shares fell by more than 2 per cent in pre-market trading in the United States (Nikkei Asia, Evertiq).

A class reading. Artificial intelligence does not run "in the cloud". Every model runs on the memory chips, and especially on high-bandwidth memory (HBM), produced by workers on shifts in clean rooms. Artificial-intelligence demand has brought memory makers record profit and record margins. The Micron workers' demand aims at exactly this point: not a one-off bonus, a permanent mechanism. The distinction matters. A bonus is the boss's favour; he decides when it will be given, and how much. Permanent sharing is a right; it becomes a subject of bargaining, and it is written into the contract. The company's announcing the reward unilaterally while mediation was still going on is a show of force about whose hands the decision will stay in.

There is this too: the shares falling with the news shows what the market knows. One of the most critical links in the supply chain is in the worker's hands. The whole valuation of the artificial-intelligence economy depends on a few thousand workers in a factory not stopping work. We discussed this material reality underneath the "cloud" myth in the piece we wrote on İlker Kalaycı's text.

Why does it matter for us? In Turkey the artificial-intelligence discussion runs almost entirely through "software" and "the user". The Micron example shows that artificial intelligence is a material industry, that this industry has workers, and that the organised power of those workers can shake the whole chain. As we told in The Revolt of Crystallized Labor, the models in which mental labour crystallises run on the chips that manual labour produces. Without an alliance of the two labours, social control over artificial intelligence cannot be established. We will follow the Taipei rally on 19 October.

2. DNB Is Removing 400 Posts: The Person Who Writes Code Is a Target Too

6 October · Norway · Artificial-intelligence agents

What happened? Norway's largest bank, DNB, announced in a statement on 6 October that it would remove about 400 full-time posts in its Technology and Services unit. The ground is openly artificial intelligence: the bank says it has put artificial-intelligence agents, which have taken over work previously done by hand, into operation in many areas, and that it has seen a serious efficiency gain especially in three areas: the checking of customer data, know-your-customer (KYC) operations, and technology development and code writing. Chief executive Kjerstin Braathen said, "Artificial intelligence is changing the way we work and the way we serve our customers." The reduction is expected to be completed in the last quarter of 2026, and the full cost effect to be seen from the second quarter of 2027. After the news, DNB shares rose by about 1 per cent (Reuters report).

A class reading. What makes this news the most instructive example of the week is that it is written in the company's own mouth. Here the question "will artificial intelligence take the jobs?" is not a guess. It is an investor-relations text. And inside the text two sentences stand side by side: 400 people are being laid off, the share is rising. The rise in efficiency is shared not as a wage rise or a shortening of working time, but as a shrinking of the payroll and a gain for the shareholder. We discussed this earlier, through Meta's "4 AM" operation, as the liquidation of the worker with their own data; and in our piece The Useless Class, the global scale of this tendency.

That "code writing" is on the list matters separately. The wave of automation from which the technology worker long saw themselves as outside has now arrived directly at their desk. In The Tearing of the Illusion we spoke of the illusion by which the technology worker sees themselves as a "labour aristocracy"; the DNB statement shows the material end of that illusion.

Why does it matter for us? In Turkey the banking and computing sectors will live through the artificial-intelligence transformation at the same speed; and without a strong representation of employees and a social-security net of the kind there is in Norway. Banking and computing workers need to talk, from today, about this demand: Let the efficiency gain that comes from artificial intelligence be shared not as job loss, but as a shortening of working time and a rise in wages. Until this demand is written into the contract, every "efficiency" statement should be read as a layoff announcement.

3. New Layoffs at Amazon: Record Investment, a Shortening Staff

7–8 October · United States, India, United Kingdom · Employment

What happened? As soon as the big discount days of October (Prime Big Deal Days) ended, Amazon started a new round of layoffs in its retail unit. According to Reuters and Business Insider, fewer than 1,000 white-collar workers were laid off. Among the units affected, besides the retail unit that runs the main e-commerce site, are customer service and seller services; workers in the United States, India and the United Kingdom are affected. A company spokesperson said, "We have organised some parts of our Stores business so that they support our priorities better." This is a new link after the layoff of 30,000 people in 2025 and of about 16,000 in January 2026.

A class reading. Let us be honest first: Amazon did not tie this round directly to artificial intelligence, and there is no evidence, at the level of the company, on which we could say "an artificial-intelligence system took this job over". But class analysis looks not at the ground of each decision one by one, but at where capital is flowing. In the same period Amazon is planning an investment of 21 billion dollars in artificial-intelligence infrastructure in India alone, up to 2030 (see section 8). Money is flowing to the data centre, to the machine, to fixed capital; the share set aside for human labour is narrowing. The tendency Marx called the rising organic composition of capital is happening today under the name "artificial-intelligence investment".

The timing of the layoff is a message on its own: meet the busiest sales days of the year together with the workers, then show them the door.

Why does it matter for us? In the discussion of "artificial-intelligence layoffs" we have to avoid two mistakes. The first is to tie every layoff to artificial intelligence; this serves, without meaning to, the companies' narrative that "technology is inevitable". We discussed the function of this narrative in What Does the "AI Tsunami" Conceal?. The second is to ignore the relation between investment and employment because a direct link with artificial intelligence cannot be proved. The right question is this: In whose hands are investment decisions taken, and to whose benefit?

II. On the Organising Front: From the White Collar to the Platform Worker

4. The Union Debate in Silicon Valley Is Gathering Strength Again

11 October · United States · Unionising

What happened? A reported piece published in KQED under Ella Jackson's byline examines how Silicon Valley workers, who long stayed away from the union, are turning to organising because of artificial intelligence, job security, and the companies' political relations. A few figures from the piece:

According to the US Bureau of Labor Statistics, only 3.7 per cent of workers in computer and mathematical occupations are unionised; in other sectors the rate is 10 per cent.
According to the Stanford Artificial Intelligence Index for 2026, entry-level software-developer employment has fallen by 20 per cent since 2024.
The Alphabet Workers Union has "more than a thousand" members in a company of about 190,000 workers.
The campaign Google workers started against relations with the immigration police, ICE, reached 1,700 signatures.
Wikimedia workers decided to unionise in September; at Blizzard Entertainment a union contract was approved. In both processes the use of artificial intelligence was on the agenda.

Joshua Carroll, a Google product manager from the Tech Workers Coalition, tells how his own job has turned "from making product decisions to shovelling data into artificial-intelligence systems". Jeremy Bernick of the CWA draws attention to workers' use of artificial intelligence being measured, in performance reviews, by "token tracking". According to the historian Margaret O'Mara, the Valley's managers were "quite harshly anti-union" from the start; stock options were also a way of offering a benefit without a union. In July Arielle Pardes in The Guardian reported the same tendency, with the examples of Meta, Google DeepMind and ZeniMax.

A class reading. There are two important findings in this piece. The first is the function of the stock option as a class instrument. Giving the worker a small share of property makes them a party not to their own labour, but to the company's valuation. This is the material basis of the technology worker feeling like an "entrepreneur" or a "partner". The second is "token tracking." Measuring the use of artificial intelligence and turning it into a performance criterion is the plainest form of what we called mental Taylorism in our book Labor in the Age of Artificial Intelligence: the process of thinking is broken into parts, measured and supervised. Last week we discussed the legal limit of surveillance at the workplace, in the Turkish context, through the KVKK's email decision.

The sceptical voice in the piece is worth attention too: there are many workers on an H-1B visa who carry a fear of retaliation. This shows that the obstacle in front of organising is not only ideological, but legal and material as well. The green-card decision we will see below in section 10 makes plain why this fear is not groundless.

Why does it matter for us? In Turkey computing workers are left, in practice, unorganised, because of the 1 per cent branch threshold and because computing is not recognised as a separate branch of industry. As we discussed in detail in The Computing Worker's Handbook, the rate of 3.7 per cent in the United States is not far from us; the rate here is probably even lower. And the problem is not only the lack of organisation: as we saw in September in the statement made at DeFacto's door for the reinstatement of four computing workers, in Turkey technology jobs are being made precarious by subcontracting. The news from Silicon Valley is this: the period of "I get a good salary, I don't need a union" is closing. In the words of John Chadfield, one of the founders of UTAW, the technology workers' union in Britain: "You may be well paid, but you are still a worker."

5. Tens of Thousands of Platform Drivers in a Union in California

10–11 October · United States · Platform workers

What happened? The Wall Street Journal's report of 10 October, "America's Gig Workers Turn to an Old-School Union", reports that tens of thousands of Uber and Lyft drivers in California have joined together in a union. The background is this: a law passed in California in 2025 recognised a right of collective bargaining for platform drivers despite their "independent contractor" status (Fortune). The California Gig Workers Union (CGWU), supported by the SEIU, passed in August the threshold set by PERB, the state's public employment-relations board; that is, it organised by signature at least 30 per cent of the active drivers in the state (CalMatters). About 350,000 ride-share drivers work in the state. The union aims to sit at the bargaining table with Uber and Lyft by the end of the year. On the bargaining agenda, besides low wages, are being thrown off the application (account closure) and the effects of the companies' algorithms on the work. In the words of Riko Mendez of SEIU 521, the drivers also want a say over "the policies and the use of the technology that shapes their jobs" (SEIU 521).

A class reading. The historical importance of this organising is that algorithmic management is being put on the bargaining table for the first time at this scale. The basic trick of platform capitalism was to appear as if it had abolished the boss: no chief, no manager; only "the app". But the app is the boss itself. It distributes the job, it sets the wage, it cuts the penalty, it closes the account. In our piece An AI Boss Fired a Human for the First Time we discussed the next stage of this relation. When drivers ask for a right to object to account-closure decisions, what they are actually saying is this: The algorithm is not a neutral law of nature. It is a management decision; and a management decision is objected to.

We should also enter a critical note here: the California model gives a limited bargaining right without recognising the drivers' status as workers. This is a gain, but it also carries the risk of fixing the "independent contractor" fiction in place. Criticisms arguing that this deal could weaken the drivers were also published in the left press (Jacobin). Owning the gain and seeing its limits can be done at the same time.

Why does it matter for us? In Turkey couriers work under the same algorithmic management, in much heavier conditions; they are pushed outside the rights of labour by the "owner-operator courier" status. There is now a concrete example in the world of the demands we discussed in the Gig Economy Dossier, especially the demand for a right to object to algorithmic decisions and for account closure to be tied to a safeguard against arbitrariness. Following the bargaining process in California will be directly instructive for courier organising in Turkey.

6. Wizards of the Coast Workers: A Binding Rule on Generative Artificial Intelligence

7–8 October · United States · Creative labour

What happened? The teams working on Dungeons & Dragons, Magic: The Gathering and Duel Masters at Wizards of the Coast, owned by Hasbro, are joining the Wizards of the Coast-CWA union, which began in April with the Magic: The Gathering Arena team (Gamereactor). Among the demands are that management stop encouraging generative artificial intelligence and adopt an enforceable, binding policy on the subject; and that the full ownership of the creative work workers produce outside working hours stay with them. The company was given until 13 October for voluntary recognition; if it is not recognised, the union will have to apply to the US National Labor Relations Board (NLRB) and win a new election.

A class reading. These demands ask two different questions of property at the same time. The first: can the company feed the creative corpus the workers have produced over the years (drawings, texts, game designs) into generative artificial intelligence and produce a tool that will replace them? This is the workplace-scale counterpart of the process we called The Expropriation of the General Intellect. The second: to whom does the worker's creativity outside working hours belong? In the creative sectors, employment contracts increasingly tend to count the whole of the worker's mental production as company property. The union's demand draws a limit against this enclosure. We took up the relation of copyright and property in detail in The Dialectic of Copyright in the Digital Age.

Why does it matter for us? In Turkey the games sector is a field that is growing fast and is almost not organised at all. Young creative workers in games, drawing, voice acting, translation and content production are among the first targets of generative artificial intelligence. The Wizards example shows that the demand of these workers can be not "ban artificial intelligence", but "let us decide together how artificial intelligence will be used, and let this decision be written into the contract". In A Debate on Writing with Artificial Intelligence we questioned not the tool itself, but the relation of use.

7. The Workers Who Run the Electricity Market Are Extending the Action

30 September–13 October · Australia · Energy and critical infrastructure

What happened? The Australian Energy Market Operator (AEMO) is the not-for-profit institution that runs the country's electricity grid and energy market. When collective-agreement talks jammed, members of the Professionals Australia union, which organises engineers and technology specialists, have been carrying out six different protected industrial actions since 30 September. According to AEMO's statement of 12 October, union members will be able to take seven new actions from 13 October. While the first actions were limited to meetings, communication and administrative processes, the new actions cover operational and technical work such as grid-outage operations, support to market participants, connection applications and software change management. The institution's second union, the Australian Services Union (ASU), has also started an action-ballot process at the Fair Work Commission. AEMO announced that it could resort to legal means "if a risk arises for public health, safety or the economy".

A class reading. This action makes visible the place of technology labour in critical infrastructure. The electricity grid is now run with software, data and algorithms; when the engineers and specialists who keep these systems standing extend their action as far as "software change management", they remind us of the labour behind society's most basic need. Management's answer is familiar too: the worker's action is presented as "a cost to the consumer" and "a risk to public safety". But the right to determine working conditions in an institution that serves the public is the public's own right as well. We also pointed out, in Anthropic's Activist Monitoring System, how the concept of "critical infrastructure" is used against labour actions.

A note for context: in a submission to the Australian parliament in March 2026, the ASU proposed putting the four-day working week into national labour standards, in a period when companies are replacing workers with artificial intelligence (ACS Information Age). The demand that the efficiency gain brought by artificial intelligence be shared by a shortening of working time is on the agenda of the Australian union movement.

Why does it matter for us? In Turkey energy, telecommunications and public computing are the fields where organising is hardest, and where organised labour could be strongest. Engineers and technical staff working in these fields are systematically kept outside union rights, by strike bans and by the ground of "critical infrastructure". The AEMO example shows two things: technical specialists can take action too, and the effect of the action comes precisely from the critical character of the work.

III. In Place and in Society: The Cost of Infrastructure and of Surveillance

8. Popular Objection in India to Google and Amazon Data Centres

9 October · India · Ecology and infrastructure

What happened? According to Reuters's extensive report of 9 October, more than 1,000 residents of the area are saying that the 54-acre hyperscale data centre Amazon plans in Thane, near Mumbai, was imposed without a meaningful consultation. The project is part of the 21-billion-dollar artificial-intelligence infrastructure investment Amazon plans to make in India up to 2030. Trees were cut to open the land; on 8 September hundreds of people, including the elderly, gathered beside the site (The Wire). The banners read "Data centres are energy vampires." In the words of Sadhana Pradhan of the "Wake Up Thanekar" campaign: "Half of Thane lives on tanker water." (Let's Data Science).

In the city of Visakhapatnam, in the state of Andhra Pradesh, Google's data-centre project of 15 billion dollars is meeting a reaction on the grounds of water scarcity, the effect on wildlife, and the insufficiency of the environmental assessment (The CSR Journal). According to local figures, about 410 million litres of water a day are supplied to the city, while estimated demand is 480 million litres. The protesters' slogan is plain: "We cannot drink data."

The figures show the scale of the problem: India holds 18 per cent of the world's population and only 4 per cent of its freshwater resources. S&P Global estimates that 60–80 per cent of the data centres in India may face high water stress within this decade. Amazon and Google say their projects include measures that limit the local effect, and that waterless cooling and a power line separate from housing will be used. The Indian government, meanwhile, grants data centres infrastructure status, and a tax holiday until 2047 to foreign cloud providers. A detailed account of the reactions can also be followed on the Business & Human Rights Resource Centre page.

A class reading. There is a classic capitalist division of labour here: the profit is private, the cost is social. The data centre's income is the company's; but the electricity grid, the water basin, the land, the noise and the heat load are the neighbourhood's. The state is the intermediary of this division of labour: with a tax holiday, infrastructure status and the discourse of "digital development" it opens the way for capital, and leaves the neighbourhood outside the decision. The residents' emphasis matters too: their objection is not to artificial intelligence itself, but to the local effect and the exclusion. This is not opposition to technology. It is a class question about for whom the resources will be used. We discussed the material metabolism of the "cloud", with the concept of "metabolic rift", in our piece on İlker Kalaycı's text.

Let us note this too: the general manager of a data-centre operator in India says he would never build a data centre in a residential area, because there is a risk of being left without electricity and water there. So even from inside capital there are people who say "this place is wrong"; the decision is still taken by the logic of competition and of incentives.

Why does it matter for us? Turkey is entering the data-centre race too: Türksat's large data centre in Gölbaşı and private-sector investments are on the agenda. Drought and water stress are Turkey's reality as well. One of the urgent demands we listed for Turkey in the Socialist Artificial Intelligence Manifesto was exactly this: Let the water and energy budget of every data-centre project be made public; let local government and local people be recognised a right of veto. We discussed how this demand can be put into practice at the local level in our guide Joining Local Government. Those who defend the forest at Akbelen and those who defend their water in Thane stand on different fronts of the same class struggle.

9. Flock Safety: Backlash at a Surveillance Company, and Layoffs

8–9 October · United States · Algorithmic surveillance

What happened? The US surveillance company Flock Safety, known for its artificial-intelligence licence-plate-reading cameras, is preparing, according to a Reuters exclusive, to lay off 18 per cent of its roughly 1,500 workers, that is 270 people. This comes after the voluntary-departure programme in September. The company operates about 120,000 cameras in 49 states, and serves more than 4,800 law-enforcement agencies and about a thousand businesses. In the latest investment round, led by Andreessen Horowitz, it was valued at 7.5 billion dollars. But the company is now facing a serious social backlash: according to a Reuters/Ipsos poll, 47 per cent of Americans are against Flock cameras in their neighbourhoods. In September Florida banned automatic licence-plate readers on state highways, on the ground of privacy risks. After reports that Flock data was being used in immigration enforcement, Home Depot investors asked for a review of relations with surveillance suppliers. A lawsuit filed in Virginia claims that the company's widespread camera network is "surveillance without a search warrant".

A class reading. This news shows two things at once. The first is that social opposition works. Surveillance technology is not "inevitable"; it can be pushed back in neighbourhood meetings, in state legislatures, in the courts and at shareholder meetings. The Florida ban and the majority opposition in public opinion forced a company created by venture capital of nearly a billion dollars to shrink.

The second, and we should not ignore this either: the bill for the backlash is cut to the worker first. It was not the workers who decided the policy of the surveillance network, but the managers and the investors. But when the company shrank, it was 270 people who were shown the door. The struggle against surveillance and the technology workers' struggle for job security are not enemies for this reason. They should be allies. The workers collecting signatures at Google against Project Maven, and today against relations with ICE (section 4), show that this alliance is possible.

Why does it matter for us? In Turkey licence-plate reading, face recognition and city cameras are spreading fast, with almost no public discussion and no oversight. In Anthropic's Activist Monitoring System we discussed how the target list of the surveillance apparatus widens (terrorist → migrant → dissident → worker). In the Flock example too, the claim that the data has been carried over to the hunt for migrants is exactly one link of this widening. We took up how surveillance is built by the hand of a private company in The Palantir Manifesto, and how it seeps into everyday life last week in Whose Eye Are the Glasses?. Our demand is plain: Let the contracts, the data-sharing protocols and the usage statistics of public surveillance systems be open to the public.

10. The Green Card Suspended: Dividing the Worker in the Name of "Protecting the American Worker"

8 October · United States · Migrant technology labour

What happened? At a press conference at the White House on 8 October, US Vice President JD Vance announced that Microsoft, together with computing companies such as Wipro, Infosys, Tata, HCL, Cognizant and Capgemini, had been suspended from the PERM programme, the first step of the green-card process for foreign workers (The Next Web, New York Times). Vance accused the companies of fraud, and said that last year Microsoft received approval for more than 6,000 H-1B visas while laying off about 6,000 American workers; he claimed that for every worker laid off, "one and a half foreign contracted slaves" were taken on. Microsoft objected to this claim, announcing that 80 per cent of H-1B applications were for people already working at the company. The same week the Department of Homeland Security proposed a fee of 70,000 dollars for foreign students' post-graduation work permit (OPT) (OnLabor).

A class reading. The visible language of this decision is friendly to the worker: "Protect the American worker." But what is being done is to set the worker against the worker. Microsoft's and the outsourcing companies' using layoffs and the import of cheap labour together is a real strategy of exploitation; exposing it is right. But the solution is not to make the visa worker even more precarious, to close the green-card path, and to keep them in a "contracted" status wholly dependent on the employer. The opposite: the more the migrant worker's dependence on the employer grows, the more that worker fears to organise, and the more convenient an instrument they become for pulling the local worker's wage down. This is what the sceptical voice in the KQED piece warned of: the fear of retaliation among H-1B workers is one of the greatest obstacles in front of organising.

The nationalist discourse of "protecting the worker" is capital's strategy of dividing the worker, applied by the hand of the state. We told this class function of racism in detail in What Is This Thing Called Racism?; and why workers vote for the right in Why Do Workers Vote for the Right?. We choose neither racism nor the companies' exploitation; we defend the equal rights of all workers.

Why does it matter for us? Turkey's computing workers stand at both ends of this picture at once: on the one hand, thousands of young programmers who go abroad, especially to the United States and Europe, are inside exactly these visa regimes; on the other, a similar nationalist discourse against migrant labour is constantly provoked in Turkey. This news is a reminder that the "brain drain" discussion is a class discussion too. We also took up the class picture of the recent period in the United States in A Class Balance Sheet of a Week in Washington.

IV. From Turkey

11. The TMMOB's Artificial-Intelligence Report: To Whose Benefit Is the Technological Transformation?

9 October · Turkey · Professional organisations

What happened? The TMMOB published the Report on Artificial Intelligence Technologies and Their Possible Effects in the Professional Fields of Engineers, Architects and City Planners. The report treats artificial intelligence not only as a technical development, but as a force that cannot be separated from capital accumulation, global competition and political power, and that reshapes working life and public services. It opens to discussion the headings of the devaluation of labour, precarity, digital control, the monopolisation of data, gender inequality, ecological effects, and the erosion of democratic rights. The TMMOB rejects both uncritical enthusiasm and wholesale opposition; it poses the basic question as "in whose interest was this technology produced, for what, and what are its social consequences?"

A class reading. We discussed the report, and the directions in which it needs to be developed, in detail in Reading and Developing the TMMOB's Artificial Intelligence Report. Its place in this bulletin is here: while this week workers all over the world are making artificial intelligence a subject of collective bargaining, in Turkey almost the only institutional voice conducting this discussion systematically is a professional organisation. This is both valuable and limited. Valuable, because the report's frame is right: technology is not neutral, and it cannot be separated from the question of property and control. Limited, because the chamber represents only the diploma-holding engineer, architect and planner; the product manager, the designer, the test specialist, the data labeller and the call-centre worker are not at this table.

Why does it matter for us? The TMMOB report is a good occasion to discuss again the division of labour between chamber and union in Turkey. The analyses in the report need to turn into clauses of a collective agreement, as at Micron, at AEMO and in California. That is possible only with the union organisation of computing workers. The chamber produces knowledge, the union bargains; as we told in The Computing Worker's Handbook, the two together are "dual power". For the historical background of this discussion, see also On the Threshold of Bilisim-Sen.

12. The Wave of Digital Censorship Silences the One Who Counts It Too: EngelliWeb Was Blocked

1–7 October · Turkey · Digital rights

What happened? An access block was imposed on 1 October, on the ground of "the protection of national security and public order", on the social-media account of EngelliWeb, the project of the Freedom of Expression Association (İFÖD) that has for years kept a record of internet access blocks in Turkey; on 5 October, X made the account invisible from Turkey (Medyascope). According to the figures the İFÖD announced on 2 October, at least 1,270 accounts were blocked in September 2026 alone; 941 of them in the second half of the month. For comparison: in the whole of 2025, the number of X accounts blocked under article 8/A of Law No. 5651 was 995. In the first week of October the accounts of journalists, academics, lawyers and news sites were added to the list too.

A class reading. We followed how this wave grew in From Obscenity to National Security and They Shut Down What They Cannot Buy; and we took up the EngelliWeb block at first hand on 2 October in They Silence the One Who Keeps the Ledger Too. This week's development has a separate meaning: the one who counts the censorship is being silenced too. A public record of the blocks was being kept not by the state, but by the voluntary labour of an association; now that record itself is being made invisible. And the decision is the state's, but the enforcement is a company's: X applying the decision in Turkey shows where the platform monopolies' discourse of "freedom of expression" ends when commercial interest is at stake. We told this partnership in Journalists' Open Call to Musk, and the Partnership of Censorship, and the technical infrastructure of blocking in How Does Blocking Work, Technically?.

Why does it matter for us? Digital censorship is a direct attack on the working class's channels of organising and of communication. If today the journalist's and the academic's account is closed, tomorrow the voice of the worker on strike, of the courier in the resistance tent, of the computing worker who has been dismissed will be closed. Keeping the record of the blocks, multiplying these records and keeping them alive on different platforms is the task of all of us. For this we should also invest in free tools such as Jitsi Meet and in our own infrastructure.

In Brief: Erdoğan's Artificial Intelligence That "Puts Labour Out of Use"

Speaking on 6 October at the 2026–2027 Academic Year Opening Ceremony at Beştepe, Erdoğan said, according to NTV, "Reaching the source and knowledge through artificial-intelligence models has also, to a large extent, put the labour in the research process out of use," and he said that "the human factor is diminishing" in academic processes.

The observation points, in itself, to a real concern. But labour being "put out of use" is happening not only in the student's homework, but in the factory, in the bank, in the call centre, in the software company. And Turkey's Artificial Intelligence Action Plan for 2026–2030 does not contain a single sentence about the millions of workers who will be left under this transformation. There is no labour representative at the table of the National Artificial Intelligence Board; the public does not know what the Directorate General of Public Artificial Intelligence is doing; and the discourse of "indigenous and national" artificial intelligence covers imperialist integration. If the concern is real, the first step is clear: labour representation on the National Artificial Intelligence Board, ratification of the ILO's convention on platform work, and recognition of computing as a separate branch of industry.

The Common Class Meaning of These Developments

Twelve pieces of news come together on four basic axes of conflict.

The first axis: the sharing of the productivity gain

In the DNB and Amazon examples, the reduction of the workforce and technological investment move forward in the same period; at DNB the share rises with the layoff news. The Micron workers, meanwhile, want a permanent share of the profit that has been created. The issue is which of three possible forms the efficiency gain brought by artificial intelligence will be shared in: a gain for the shareholder, a rise in wages, or a shortening of working time. Capital chooses the first; labour's demand is the second and the third. While saying this, let us repeat a warning too: to say that every layoff is a direct substitution by artificial intelligence, evidence at the level of the company is needed. In the Amazon example this evidence does not exist; in the DNB example the company itself says it. We made the finding that "whose the diminishing time remains with is a question of property" in our piece on Bill Gates's warnings; this week filled up with concrete examples of that finding.

The second axis: the authority to decide at the workplace

The drivers in California want a right to object to algorithmic account-closure decisions, the Wizards of the Coast workers a binding rule on the use of generative artificial intelligence, the specialists at AEMO a collective agreement; in Silicon Valley "token tracking" is being discussed. The discussion is no longer only the wage: Who decides how the technology will be applied? This is the most current form of democracy at the workplace. We discussed the companies' claim to answer this question among themselves in Can Companies Audit Themselves? and Whose Hand Is on the Shutdown Button?.

The third axis: the social cost of infrastructure

Data centres are the property of private capital, but their loads on the electricity grid, water resources, land and the environment are distributed to society. The objections in India make this problem of distribution visible. The infrastructure of artificial intelligence that is presented as "the cloud" in fact stands in someone's neighbourhood, beside someone's water well. We showed, in Artificial Intelligence Bosses Are Planning the Anger, Not the Disaster, that the cost of the disaster scenarios is being prepared, by the same logic, to be cut to society.

The fourth axis: the role of the state, surveillance and division

Perhaps the least talked-about axis of this week is the class function of the state in the field of technology. In India the state opens the way for capital with a tax holiday; in the United States it makes the migrant worker even more precarious in the name of "protecting the American worker"; in the Flock example, law enforcement is the largest customer of a private company's surveillance network; in Turkey the state gives the access-block decision, and the platform monopoly applies it. The state is not a neutral referee in this field; it is the organiser of capital's technological domination. We also developed this frame in The Declaration of Techno-Fascism and The Same Week, Two Tables.

For this reason the class agenda of technology in the recent period should not be reduced to the question "Will artificial intelligence take our jobs?" The more comprehensive question is this: How will the relation between the ownership of the means of production and the democratic control of technological decisions be established? Our answer to this question can be found, gathered together, in the Socialist Artificial Intelligence Manifesto and in The Communist Manifesto 2.0.

What We Will Follow Next Week

Date	Development	Why are we following it?
13 October	The end of the period for voluntary union recognition at Wizards of the Coast	Will artificial-intelligence rules in creative labour enter the contract?
From 13 October	Extended actions at AEMO	The bargaining power of technology labour in critical infrastructure
19 October	The Micron union's rally in Taipei	The process leading to a strike date; labour in the artificial-intelligence supply chain
Ongoing	CGWU–Uber/Lyft bargaining preparations in California	Algorithmic management at the table for the first time at this scale
Ongoing	Account blocks in Turkey	Whether the wave of censorship is reaching worker and union accounts

Concrete Tasks: What Falls to Us

Write an artificial-intelligence clause into the contract. In workplaces in banking, computing and services, a draft collective-agreement clause should be opened to discussion, demanding that the efficiency gain from artificial intelligence be shared not as layoff but as a shortening of working time and a rise in wages.
A right to object to algorithmic decisions. For couriers and platform workers, the demand for a mechanism to object to account closure, score cuts and job-allocation decisions should be raised, with the California example as a reference (Gig Economy Dossier).
A separate branch of industry for computing. The discussion the TMMOB report opened should be joined to the demand that computing be recognised as a separate branch of industry and that the 1 per cent threshold be lifted (The Computing Worker's Handbook).
The water and energy budget of data centres. The water and energy consumption, the environmental-impact assessment and the public incentives of every data centre being built or planned in Turkey should be made public (Socialist Artificial Intelligence Manifesto).
Transparency in surveillance contracts. The licence-plate-reading, face-recognition and city-camera contracts of municipalities and public institutions, and the data-sharing protocols, should be open to the public (Joining Local Government).
Multiply the records of blocking. The records of EngelliWeb and the İFÖD reports should be shared on different platforms, and local and union media should keep these data alive.
International solidarity. The struggle of the Micron workers, of the California drivers, and of the neighbourhood in Thane that is defending its water should be carried, with messages of solidarity, onto the agenda of the unions and professional organisations in Turkey.

Closing: Not Opposition to Technology, but a Say over Technology

The developments of this week teach us something: all over the world workers are struggling not against artificial intelligence, but over in whose hands, and to whose benefit, artificial intelligence will be used. The Micron worker is refusing not the production of the chip, but the one-sided sharing of the profit. The California driver is refusing not the application, but the arbitrariness of the application. The neighbour in Thane is refusing not the data centre, but the taking of their own water.

It is labour that produces technology. The labour that casts the chip, writes the code, labels the data, keeps the grid standing, brings the parcel to the door. This labour demanding a say over how the technology it produces will be used is the most current form of the class struggle of the twenty-first century. The answer to Is a Real Emancipation of Artificial Intelligence Possible? passes through here too.

Until we meet in the new issue next Monday.

Knowledge belongs to everyone.