The National AI Board: Who Is at the Table, Who Is Not?

Whose Board Is This Board? A Class Anatomy of AI Governance

Dear Young Comrade,

Ten days ago we discussed Turkey’s Artificial Intelligence Action Plan (2026–2030) on these pages. We said the plan itself was less a development programme than an accumulation programme organised by the state against the tendency of the rate of profit to fall.

On 27 August 2026 that programme’s state apparatus was announced. According to Anadolu Agency (AA, 27 August 2026; TRT Haber), three new structures are being established: a National Artificial Intelligence Board, a Program Office, and a National Artificial Intelligence Ethics Board.

The plan had set targets. Now it is becoming clear who will implement those targets, with what authority, and answering to whom. And a programme’s class character is read most clearly precisely here — in its institutional architecture. Because targets may be wishes; but the distribution of authority is not a wish, it is the power relation itself.

So let us sharpen the question: This table has been set; who is at the table, and who is not?

Part One: What Is Being Established?

First let us set down the facts in the official language’s own wording.

1. The National Artificial Intelligence Board

Its chair is the President. According to the report, the Board will be "the highest decision-making and coordination organ with respect to strategy, prioritisation, resource allocation, and risk and ethical principles."

Its duties:

To approve national priorities and annual implementation plans
To resolve inter-institutional obstacles and eliminate overlaps of authority
To direct large-scale investments
To publish once a year a "National AI Policy Letter" — whose announced function is to guide public and investor decisions.

2. The Program Office

The technical-administrative centre of implementation. It will projectise and track the actions and operate the "National AI Progress and Transparency Portal." The portal will open progress to the public with green–yellow–red status indicators — but "while protecting personal data and commercial secrets."

Within the Office will sit the Presidency of Cyber Security’s Public AI Directorate General; coordination of AI applications in the public sector will run from there.

3. The National Artificial Intelligence Ethics Board

It is targeted to be established within one year. Its composition: representatives of the public sector, the private sector, academia, and civil society. Its task: to issue guidance and evaluate applications on justice, bias, transparency, security, and human oversight. Children and vulnerable groups are marked as the priority field.

This will be accompanied by a task matrix showing, for every action, the responsible institution, the decision-maker, the implementer, the technical overseer, and the budget allocator.

On paper, a tidy architecture of governance. Now let us look at this architecture with a class eye.

Part Two: Class Assessment

1. Who is not at the table?

If you remember a single sentence of this piece, let it be this: Labor is not on the highest decision-making organ.

The Board convenes under the chairmanship of the President, and its function is resource allocation and the directing of large investments. That is, this is the organ that will decide, over the next five years, which job in which sector will be automated how, where which data centre will be built, which public service will be handed over to an algorithm.

The class directly addressed by these decisions — the call-centre worker, the hospital secretary, the driver, the courier, the accounts clerk, the public employee, the software worker — is not represented at this table.

Unions, professional organisations, workers’ confederations are absent. There is no heading that speaks of labor.

The names of labor organisations can dissolve, at most, inside the phrase "civil society" that appears in the composition of the Ethics Board. And the division of labour here is highly instructive:

Decision-making organ: the state and capital. Advisory organ: "society." And the advisory organ will be established a year later.

While investment decisions begin to be taken from today, the ethics-evaluation mechanism comes a year behind. This is not a hitch; it is a sequence. Capital moves first, ethics catches up later; and when it catches up, it evaluates not the tableau in front of it, but the wreckage that remains.

2. "Ethics" produces not sanction but legitimacy

The Ethics Board’s powers must be read carefully: to issue guidance and to evaluate applications.

What is missing from both of these is a single thing: binding force.

The situation of a worker whose performance is measured by an algorithm, whose shift is assigned by a model, or who is dismissed directly by a software output is not solved by a "guide." If an ethics guide does not generate a sanction when it is violated, it is a text whose cost to the firm is zero. A zero-cost rule is not a rule; it is an announced intention.

The ideological side of the matter is here too: The language of ethics turns a class conflict into a technical quality problem. "Bias," "transparency," "human oversight" — all are real headings. But a worker’s losing their job is not a "bias" problem; it is a problem of property and power. Even if the algorithm runs impartially, that worker loses the job.

The task of ethics is precisely to make this distinction invisible: The system itself is not debated; the system’s "responsible use" is debated.

3. A single centre: producing predictability for capital

One of the Board’s most striking duties is this: "to resolve inter-institutional obstacles."

Removing the fragmentation inside the state’s own bureaucracy looks technically reasonable. But let us ask what it means in class terms: Whose encountered obstacle is being removed?

That of the large-scale investor. The energy permit, land allocation, environmental assessment, data access, and regulatory exemption of the capital that will build a data centre — including the "regulatory sandboxes" mentioned in the Action Plan — will all be resolved at a single table, at the highest level.

Bureaucratic friction is a cost for capital. This Board is an apparatus that lowers that cost. The message given to the investor is clear: Your decision will be taken here, quickly, and at the highest level.

The classic proposition of Marxist state theory here ceases to be a theoretical abstraction and turns into an organisational chart: The executive of the modern state is but a committee for managing the common affairs of the whole bourgeoisie. That is the content lying under the phrase "coordination organ."

Let us clarify one point here, because the quality of the debate depends on it: Our problem is not planning as such. On the contrary, as we said in the previous piece, that the social planning capitalism rejects as "incalculable" has today become technically possible is a historical irony, and it is in our favour. Our problem is in whose interest and under whose control the planning is done. The same computational power can organise social need, and it can organise the rate of profit.

4. The transparency portal: transparency of what?

The "National AI Progress and Transparency Portal" will show progress with green–yellow–red lights. It sounds like accountability.

But the object of the transparency is the progress of the action plan, and its subject is public institutions. That is, the portal answers the question "was the target reached?" Not the question "who won, who lost?"

Moreover there is a reservation in the announcement: the data will be shared "while protecting personal data and commercial secrets." The commercial secret reservation is wide enough to cover everything we actually want to know.

The transparency a worker really needs is this:

Which algorithmic system is used in the workplace where I work?
On what data was this system trained; was my data used?
By what criteria does it evaluate me?
After this system was brought in, how many people were dismissed, how many people’s wages fell?
Who is the appeal authority — a human or a machine?

None of these will appear on the coloured dashboard. The transparency of governance and the transparency of production relations are not the same thing. The first is a shop window; the second is a matter of struggle.

5. Who is the addressee of the "Policy Letter"?

The function of the Policy Letter the Board will publish once a year is written out clearly: to guide public and investor decisions.

Pay attention to the list of addressees in the sentence: the public and the investor. No worker, no union, no consumer, no neighbourhood, no villager who will share the data centre’s water and electricity.

This is not a slip of the tongue; it is a correct identification of the target audience. Here the state is giving capital an annual risk-reduction signal: Enter these fields, you will be supported here, we will not step back from here.

For capital, uncertainty is the most expensive input. This letter cheapens that input.

6. Placing public AI inside the security apparatus

One of the least discussed but most important details: the positioning of the Public AI Directorate General inside the Presidency of Cyber Security.

The coordination of AI applications in the public sector is being organised not in a social-policy institution, a development agency, or a public-service unit, but in the security apparatus.

This shows in which frame public AI is conceived: first a security matter, then a service matter. If the architecture of systems to be built in health, education, and social assistance is shaped from this frame, the citizen is constructed first not as a recipient of service, but as a subject watched and given a risk score.

The issue we will discuss under the heading of the digital panopticon in our Karaburun presentation is precisely this — and note: this is not merely a privacy problem. Surveillance is first of all a technology for overseeing and disciplining the labor process; the privacy violation is its by-product.

7. "Children and vulnerable groups" — and what about workers?

That children and vulnerable groups are marked as the Ethics Board’s priority field is not something to be opposed. It is a real and serious need.

But this correct heading is precisely what makes the absence of another heading visible: workers.

Because when the protection frame is built through "vulnerability," those protected are defined as victims. What the victim needs is patronage. Yet the working class is not a stratum to be patronised; it is a class that demands rights and has bargaining power. You cannot put labor under the heading "vulnerable group"; because labor’s problem is not a lack of sensitivity, it is exploitation.

The limit of the language of ethics is visible here: It can produce mercy; it cannot change the balance of power.

8. Gates’s global board, Ankara’s national board

In our previous piece we discussed Bill Gates’s manifesto of 26 August 2026. There Gates proposed a global AI institution inspired by nuclear-oversight and civil-aviation regimes — and wanted this institution founded by "the countries that host the leading AI developers and control critical parts of the supply chain."

That the national board was announced from Ankara a day later is not a coincidence; it is two ends of the same conjuncture. The language of governance is being built at the same time on a world scale: board, office, ethics committee, progress indicator, transparency portal.

But the two tables are not equal. The table Gates describes is set by those who hold the model, the chip, and the cloud infrastructure. Ankara’s table, as we showed in the previous piece, is being set on a ground dependent on a quota regime in access to advanced AI chips, on imports in energy, on global platforms in the software stack.

In this case the national board’s objective function is not to found sovereignty, but to manage dependence. The distance between the discourse of "digital sovereignty" and who sells the data centre’s chip is this programme’s real limit. Sovereignty is measured not by the geographic location of the server, but by property in the means of production.

9. Are we then against the board?

No. And let us make this distinction clearly, because turning criticism into rejection is the easiest error.

That the social effects of AI be taken up by a central public will is a correct need. Leaving the market to itself is worse. Tearing out the machine is not our programme; taking control of the machine is our programme.

Our objection is not to the board’s existence, but to its composition and distribution of authority:

Why is there no labor representation on the decision-making organ?
Why are the ethics organ’s decisions not binding?
Why does transparency not cover the workplace?
Why does a model trained on public data not become public property?

These are not a list of rejections; they are a list of demands. And demands become demands only when defended by an organised force; otherwise they remain wishes.

10. So who will carry this demand? BMO and Bilişim-Sen must be at the table

Writing a list of demands is easy. The real question is this: Who is the organised subject that will stand behind this demand?

In this country that subject has two concrete addresses, and neither is at the table today:

The Chamber of Computer Engineers (BMO) — a professional organisation affiliated to TMMOB, whose public-interest duty is defined by law. The professional counterpart of every technical heading the Board speaks of — model architecture, data governance, security audit, algorithmic impact assessment — lies in this chamber’s field.

The Union of Informatics Laborers (Bilişim-Sen) — founded in 2026, the union of the sector’s own workers. Organising against precarity, long hours, and low wages, this union is the concrete, not abstract, side of the "labor in the age of AI" debate: the organisation of the laborers who train the models, build the data pipeline, and keep the system standing.

These two organisations’ absence from this table is not a gap; it is an exclusion. But detecting the exclusion is not enough; it must be demanded. And let us speak plainly here: A seat is not offered to anyone of its own accord. If professional chambers and unions wait saying "we were not invited," that table is set without them, and once it is set, widening it becomes much harder.

What can be done concretely:

A formal and public demand for membership. BMO and Bilişim-Sen should convey in writing to the Ministry of Industry and Technology and the Program Office a demand for representation both on the National AI Board and on the Ethics Board — and should announce this application to the public. Even if it is rejected, a record of the rejection remains; that record is the basis of the coming struggle.
Using the one-year window. That the Ethics Board will be established "within one year" means its composition has not yet been determined. This is the only period in which the composition debate is open. After this window closes, the debate turns into one of "entering an already established structure later," which is much harder.
A joint labor platform. BMO and Bilişim-Sen are weak alone; the founding of a joint "AI and Labor Platform" with TMMOB’s related chambers (Electrical, Mechanical, Surveying), with workers’ and public-employees’ confederations, with medical chambers and bar associations, would take the demand out of being a single petition and turn it into social pressure. Because algorithmic management is not only the informatics sector’s problem; it is every branch’s, from health to law.
A shadow report: a Counter Policy Letter. If the Board will publish a "National AI Policy Letter" once a year, labor organisations should also publish a labor report once a year: which system was brought in in which branch, how many people lost their jobs, what happened to wages, in which workplaces algorithmic surveillance is applied. The state’s indicator asks "was the target reached?"; ours will ask "who won, who lost?"
Watching the Transparency Portal. Once the portal opens, following with right-to-information requests the real questions behind the green–yellow–red indicators — budget, tender, data-sharing agreements — is the most concrete public-oversight work a chamber with technical capacity can do.
Turning expertise into bargaining power. The technical knowledge the Board needs sits largely with the members of these two organisations. This is not an expectation of favour; it is a lever: Labor organisations can themselves prepare public-oversight standards, methods of algorithmic impact assessment, and the technical specification of a workplace algorithm register, and put them on the table. The rejection of a ready and competent text is far more costly than the rejection of a demand.

And a warning, because this is the easy error of this debate: Sitting at the table does not of itself change the table’s character. The risk that labor organisations taken onto advisory organs become an ornament of legitimacy for decisions already taken is real. The safeguard against this risk is also clear:

The representative’s accountability to the member and regular accounting,
The announcement from the outset of the right to dissent in public from Board decisions,
And most important: that the source of power at the table is in the workplace. Representation in a non-unionised sector is courtesy; representation in an organised sector is bargaining.

So the struggle runs on two arms: on one side the demand for institutional representation, on the other workplace-by-workplace organising. Without the second, the first is decoration.

Two Readings Side by Side

Official Discourse	Materialist Reality
"The highest decision-making and coordination organ"	The gathering of capital allocation in a single centre; labor’s absence from the table
"Resolving inter-institutional obstacles"	Reducing the regulatory friction in front of the large investor
"The Ethics Board will be established within one year"	Investment today, ethics tomorrow: the sequence itself is a priority
"Ethics guides will be published"	A norm without sanction is a zero-cost text for the firm
"Transparency Portal"	Transparency of target progress; production relations are a "commercial secret"
"The Policy Letter will guide the public and the investor"	No worker or union on the list of addressees — this is a declaration of preference
Public AI sits in the Presidency of Cyber Security	Not a public-service frame, but a security and control frame
"Children and vulnerable groups are the priority"	The language of protection produces patronage; labor wants bargaining power
"Digital sovereignty"	Chip quotas, energy imports, and platform dependence continue
"Human-centred AI"	Which human? The one who holds the share, or the one who holds the shift?
Composition "public, private sector, academia, civil society"	The professional chamber and the union are dissolved into "civil society"; a seat not demanded is not given

Concrete Demands

Criticism is commentary until it turns into a demand. Headings that can be put on the table today from this news:

On the Board’s composition and powers

Labor representation on the National AI Board. Representation of, first of all, the Chamber of Computer Engineers and the Union of Informatics Laborers, and of workers’ and public-employees’ confederations and professional chambers (including TMMOB, medical chambers, and bar associations) on the decision-making organ — not on the advisory organ.
The binding force of Ethics Board decisions and administrative sanction powers in case of violation; also an individual and union right of application.
Pulling the establishment of the Ethics Board not to a year, but to before the first investment decision.
Opening the task matrix to the public: who took which action’s decision, which budget was allocated to whom, must be trackable.

On transparency and the workplace

A workplace algorithm register. The recording of every system used in employment, performance, shifts, and pay, and its being open to the union.
Placing employment algorithms in the "high-risk" class — we defended this demand in the previous piece as well; it should be the Board’s first agenda item.
Narrowing the "commercial secret" exception: the worker’s right to learn the reason for an algorithmic decision taken about them cannot be limited by commercial secrecy.
An automation impact notice: measuring the employment impact before a new system is brought in, notifying the union, and negotiating it.

On property and the public

Models trained on public data are public property. Every model produced from the 2,000 public datasets opened should be public, together with its weights; exclusive licences should not be given to a private company.
The public employee’s tacit knowledge is not free raw material. If the professional experience of the doctor, the judge, the teacher, the civil servant is to be fed into model training, professional organisations set the rule.
Public oversight of data centres’ energy, water, and land effects; participation of local people and labor organisations in decision processes.
Informatics laborers’ organised right of refusal: a guarantee of being able to refuse to work on systems for surveillance, war, and rights violation.

It Is Not Enough to Found the Board; the Table Must Be Shared

Dear Young Comrade,

A technology’s class character becomes visible not when it is how advanced, but when it is decided at whose table.

The table set this week is a real table: it will allocate resources, direct investment, determine the form of public service. The problem is not that the table is set, but the distribution of the chairs.

And a chair is not given without being demanded. Not by an ethics guide, not by a transparency portal, not by a billionaire’s proposal for a global board. A chair is taken by organised power.

That is why our call is concrete: BMO and Bilişim-Sen should put forward, today, in writing and in public, a demand for representation both on the National AI Board and on the Ethics Board. The Ethics Board’s composition is not yet fixed; this window will close in a year. A seat not demanded does not stay empty — someone else sits in it.

We are meeting to continue this debate face to face: the 20th Karaburun Science Congress, 3–5 September 2026, Karaburun, Izmir. This year’s axis of the congress is "The Revolt of Crystallized Labor." In our presentation — "The Expropriation of the General Intellect in AI Models, Mental Taylorism, and the Digital Panopticon" — we will discuss these boards too, and those left outside them. Labor in the Age of Artificial Intelligence, our book that treats the subject as a whole, is also out.

They are founding a board to manage a transition.

We are organising to be the subject of that transition, not its object.

We want our collective intellect back.