As nations scramble to move from high-level AI principles to enforceable rules, Türkiye has emerged as a case study in the administrative turn of AI governance.

YAZEK represents a mandatory effort to embed ethics into the public sector’s workflow.

In February 2026, the Turkish Ministry of National Education (MEB) launched the Artificial Intelligence Applications Ethical Declaration System (YAZEK). While many global frameworks remain voluntary or advisory, YAZEK represents a mandatory effort to embed ethics into the public sector’s workflow.

Türkiye's move sits within a growing international conversation about AI in education. UNESCO's 2024 AI Competency Frameworks for Teachers, the Council of Europe's guidelines on AI in education, and the OECD's work on AI and learning all establish principles and literacy benchmarks, but none require individual educators to declare their use of specific tools before deployment.

Similarly, while the EU AI Act classifies certain educational AI applications as high-risk and places obligations on developers and deployers, it does not mandate a tool-by-tool declaration at the worker level. At that point, YAZEK represents an early attempt to fill that gap by operationalizing ethics at the individual level.

The Mechanism: Procedural Accountability at Scale

The YAZEK system functions as a centralized digital gateway within the Ministry’s existing administrative infrastructure (MEBBİS). Before any educator or administrator can deploy an AI tool, ranging from generative AI for lesson planning to automated grading systems, they must submit a formal ethical declaration.

The system requires users to attest to eight core principles, including Human-Centeredness, Privacy, and Transparency. By embedding these requirements into the daily bureaucracy of one of the world’s largest education systems, Türkiye is attempting to socialize ethical standards across its workforce.

For other states struggling to govern AI in public institutions, embedding declarations into existing administrative infrastructure offers one model worth examining, though its effectiveness depends heavily on the robustness of the declaration process as a whole.

Why Ethical Guidelines in Schools Matter for Policymakers

Schools are among the highest-stakes environments for AI deployment. They involve minors, compulsory participation, and consequential decisions from grading to behavioral monitoring that can shape life outcomes.

Unlike consumer-facing AI applications, where users can opt out, students and teachers rarely have that choice. Ethical guidelines in this context are not abstract aspirations; they are the mechanism by which the state signals whose interests the technology is meant to serve and what recourse exists when it causes harm. 

For policymakers and civil society looking to reboot digital democracy, YAZEK is as much a cautionary tale as it is a blueprint. It highlights a growing trend in global tech policy: the rise of self-attestation models that risk substituting procedural compliance for substantive ethical oversight.

The Critical Gap: Notification vs. Validation

Despite its structural ambition, YAZEK suffers from a flaw common in early-stage AI regulations. It confuses notification with validation.

Under the current framework, an educator can often proceed with an AI application immediately after submitting the declaration. There is no mandatory cooling-off period or external technical audit to verify that the tool actually meets the declared standards.

The system produces a digital trail of intent rather than a guarantee of safety.

In practice, this means the system produces a digital trail of intent rather than a guarantee of safety. A declaration, no matter how detailed, does not inherently lead to ethical implementation. It merely ensures that a box was checked before a black-box algorithm was introduced to a classroom.

This gap is further evidenced by the April 2026 Educational AI Teacher’s Handbook published by the Innovation and Educational Technologies General Directorate (YEĞİTEK). While the handbook provides a helpful start by listing sample tools categorized by course and skill, the burden of ethical due diligence remains firmly on the individual. 

Expecting a classroom teacher to perform an independent audit of a global tech firm's data practices represents an unrealistic expectation of technical labor.

The guide advises teachers to verify whether tools use student data for training or involve third-party data sharing. However, deciphering complex, legalistic Terms and Conditions is a task that falls outside the professional sphere of most educators. Expecting a classroom teacher to perform an independent audit of a global tech firm's data practices represents an unrealistic expectation of technical labor.

The "Remedy Gap" in Administrative AI

Perhaps the most significant challenge for democratic oversight is the system’s silence on restorative justice. While the YAZEK guidelines stipulate that an AI application must be halted if an ethical breach is detected, they provide no roadmap for addressing the harm already inflicted.

In a policy environment increasingly focused on harm mitigation, the absence of a "Right to Remedy" is a notable deficiency. If a student’s privacy is compromised or their academic record is unfairly impacted by a biased automated system, stopping the tool is only half the solution.

While the recent YEĞİTEK handbook adopts a strong normative stance, which states that all stakeholders, including developers, are responsible for potential harms, it lacks a policy-based legal mechanism for enforcement.

A robust democratic framework requires more than normative aspirations. It requires protocols for data deletion, re-evaluation of automated decisions, and transparent pathways for redress, elements that are currently missing from the Turkish model. 

Global Roadmaps: Moving Beyond the Checklist

For the global tech policy community, the Turkish experience with YAZEK suggests three critical pivots for future AI governance:

From Individual to Institutional Responsibility: Expecting individual educators to verify the transparency of global, proprietary LLMs is an unfunded mandate of technical labor. Systems like YAZEK must be paired with centralized registries of pre-vetted, high-risk tools to ensure the burden of technical verification does not fall on end users.

Bridging the Algorithmic Literacy Divide: Procedural governance is only as effective as the literacy of those navigating it. For self-declaration to be meaningful, it must be accompanied by mandatory, hands-on training that helps public servants identify the biases and privacy risks inherent in agentic AI.

Codifying Restorative Justice: A policy that only addresses "what happens if it works" is incomplete. Governments must explicitly define "what happens when it fails" and establish clear administrative procedures to restore equity to individuals harmed by automated systems.

Conclusion: Toward Substantive Oversight

Türkiye’s YAZEK system and the subsequent YEĞİTEK handbook signal that it is possible to build the infrastructure for ethical AI at a national scale. However, as AI transitions from an emerging issue to a permanent layer of governance, the focus must shift from formal compliance to substantive protection.

True algorithmic accountability requires more than a login and a checklist. It requires a system that values individual rights as much as administrative efficiency. 

True algorithmic accountability requires more than a login and a checklist. It requires a system that values individual rights as much as administrative efficiency.

As other states develop their own roadmaps, they should aim to balance the efficiency of self-attestation with the need for formal oversight and restorative measures for those affected by automated systems.