Does the EU AI Act Apply to Turkish Companies? A 2026 Guide
The EU AI Act reaches companies in Türkiye whose AI serves people in the EU. What applies when after the 2026 omnibus, chatbot rules, fines, and a 90-day plan.

If the AI assistant on an Istanbul e-commerce company's website is chatting with a customer in Germany, it has been subject to a new rule since 2 August 2026: the customer must know they are talking to an AI. It makes no difference that the company has no office, staff, or branch in the EU. The EU AI Act does not care where the provider is based, as long as the AI system's output is used in the EU.
That matters to many Turkish companies: exporters to the EU, software firms with EU customers, and any business whose website serves visitors from the EU. This guide explains when the Act applies to you, which rules apply from when after the July 2026 "digital omnibus" changes, and what to do in the next 90 days.
The short answer
- The Act applies to companies outside the EU. Turkish companies that place AI systems on the EU market, or whose AI output is used in the EU, are in scope. If you use AI only for work in Türkiye and its output is not used in the EU, the Act does not apply to you directly.
- Transparency is the most urgent duty. Since 2 August 2026, systems that interact directly with people must say they are AI unless that is obvious from the context, and deepfakes must be labelled.
- The bans have applied since February 2025 and were not postponed by the omnibus. Emotion recognition in the workplace and social scoring are two examples.
- High-risk deadlines moved. Areas such as recruitment, credit scoring, and education now apply from 2 December 2027; AI embedded in regulated products from 2 August 2028.
- Fines reach €35 million or 7% of worldwide annual turnover. For SMEs and start-ups, the lower of the two amounts applies.
- KVKK still applies. The AI Act does not replace data protection law; it sits on top of it.
What is the EU AI Act?
The EU AI Act, formally Regulation (EU) 2024/1689, entered into force on 1 August 2024 as the first comprehensive law on artificial intelligence. Because it is a regulation, it applies directly in all 27 EU member states, without each country having to pass its own law.
The Act regulates AI by its purpose and the risk it creates, not by the technology it uses. The same language model can be low-risk when it answers customer questions and high-risk when it screens job applications.
| Risk level | Examples | What is required |
|---|---|---|
| Prohibited | Social scoring, emotion recognition in the workplace and in education (except for medical or safety reasons), systems that exploit people's vulnerabilities, untargeted scraping of facial images from the internet or CCTV | Cannot be used |
| High risk | Recruitment and CV screening, promotion and performance decisions, credit scoring of individuals, life and health insurance pricing, exam assessment, AI working as a safety component in products such as machinery or medical devices | Risk management, data governance, technical documentation, record-keeping, human oversight, conformity assessment |
| Transparency | Chatbots and voice assistants, AI-generated images, audio, video, and text, deepfakes | Inform people, mark or label the content |
| Minimal risk | Spam filters, document summarisation, coding assistants, most internal productivity tools | No specific obligations |
Separate rules cover companies that develop general-purpose AI (GPAI) models, such as technical documentation, a copyright policy, and a summary of training data. The duty to take measures supporting staff AI literacy applies to all providers and deployers, whatever the risk level.
Not every system working in one of the areas listed in Annex III is automatically high-risk. Systems that only perform a narrow procedural task, or improve the result of an assessment a person has already made, may be exempt, but you must document that assessment. Systems that profile people are always treated as high-risk.
When does the Act apply to Turkish companies?
Article 2 of the Act defines scope by how the AI touches the EU, not by where the company is established. A Turkish company is in scope if any of the following applies:
- It places an AI system or general-purpose AI model on the EU market. For example, a SaaS product with AI features sold to EU businesses. Publishing a language model in a repository open to users in the EU can also count as placing it on the market.
- Its AI system's output is used in the EU. For example, a website assistant that talks to customers in Germany, France, or the Netherlands, or candidate assessments prepared for a client in the EU.
- Its company or branch established in the EU uses AI. EU subsidiaries are directly in scope as deployers.
- It exports products containing AI to the EU. Manufacturers that place a product on the market under their own name together with an AI system acting as a safety function are in scope. This concerns exporters of machinery, household appliances, automotive parts, and medical devices.
The Act defines two main roles. A provider develops an AI system, or has one developed, and places it on the market or puts it into service under its own name or trademark. A deployer uses an AI system under its authority for professional purposes. If an agency built your website assistant and it runs under your brand, you may be its provider.
Providers established outside the EU that place a high-risk AI system on the EU market must appoint an authorised representative established in the EU before doing so. The same applies to providers of general-purpose AI models.
Related: KVKK and AI: What Turkish Companies Can and Can't Send to ChatGPT
The timeline: which rule applies from when?
The Act applies in stages, not all at once. Regulation (EU) 2026/1744, published in the Official Journal on 24 July 2026 and in force since 27 July 2026, is known as the "digital omnibus". It postponed the high-risk deadlines, softened some obligations, and added new prohibitions. It did not postpone the transparency rules.
| Date | What starts to apply |
|---|---|
| 1 August 2024 | The Act enters into force |
| 2 February 2025 | Prohibited practices and AI literacy |
| 2 August 2025 | Rules for general-purpose AI models, the governance structure, and penalties |
| 27 July 2026 | The omnibus enters into force; the AI literacy duty is softened |
| 2 August 2026 | Transparency rules (Article 50) and most of the rest of the Act |
| 2 December 2026 | Machine-readable marking for generative systems placed on the market before 2 August 2026; new bans on AI that generates non-consensual intimate deepfakes or child sexual abuse material |
| 2 December 2027 | High-risk systems in Annex III, in areas such as employment, credit, insurance, education, and essential services |
| 2 August 2028 | High-risk systems in Annex I: AI embedded in products such as machinery, toys, lifts, and medical devices |
The omnibus also rewrote the AI literacy duty in Article 4. Instead of ensuring that their staff have a sufficient level of AI literacy, providers and deployers must now take measures to support its development. The duty did not disappear; only the requirement to guarantee a particular level was removed.
Transparency rules for chatbots and generative AI
For most Turkish companies, the Act's first practical effect comes from Article 50, which has applied since 2 August 2026.
- Systems that interact directly with people: A chatbot, voice assistant, or AI agent that speaks to customers on the phone must be designed to tell people they are interacting with an AI. The exception is when this is already obvious from the context.
- Systems that generate synthetic content: Providers of systems that generate audio, images, video, or text must mark the output in a machine-readable format so that it can be detected as artificially generated. For systems on the market before 2 August 2026, the deadline is 2 December 2026.
- Deepfakes: Deployers that publish artificial image, audio, or video content resembling real people, objects, places, or events, which could falsely appear authentic, must disclose that it was artificially generated or manipulated. For evidently artistic, creative, satirical, or fictional works, the disclosure can be made in a way that does not spoil the work.
- Text that informs the public: AI-generated text published to inform the public on matters of public interest must be disclosed as AI-generated, unless a person has reviewed it and someone holds editorial responsibility.
- Emotion recognition and biometric categorisation: Deployers of these systems must inform the people exposed to them.
The information must be given clearly and distinguishably, at the latest at the time of the first interaction or exposure. The European Commission published its guidelines on Article 50 in July 2026, and the code of practice on marking and labelling AI-generated content was finalised in June 2026.
A practical checklist for your website assistant:
- Say it is an AI in the opening message. A sentence such as "Hello, I'm [Company]'s AI assistant" is enough. Mentioning it only in the privacy policy or the page footer is not.
- Do not instruct the assistant to act human. It should answer honestly when asked "Am I talking to a real person?" Giving the assistant a human name and photo strengthens the impression that it is a person.
- Tell people in their language. Showing a German-speaking visitor a notice only in Turkish does not make the information clear.
- Offer a route to a person. This is not an Article 50 requirement, but it reduces complaints and prepares you for the human oversight that high-risk decisions require.
- Review AI-generated marketing images and videos. Label them if they imitate real people or events.
- Ask your vendors. Do the image, audio, or text generation tools you use mark their output with watermarks or metadata? Does the contract say who carries that obligation?
Related: Multilingual AI Assistants: The Complete Guide to Arabic, Turkish, and English
High-risk uses: HR, credit, and exported products
There is still time before the high-risk rules apply, but preparation takes long. Three situations stand out for Turkish companies:
- Recruiting for roles in the EU: AI that screens CVs, ranks candidates, or assesses interviews for jobs in the EU falls under the high-risk rules from 2 December 2027. As a deployer, you must use the system according to its instructions, assign human oversight to competent people, make sure input data is relevant to the purpose, monitor the system, keep its automatically generated logs for at least six months, and inform workers and their representatives before using it in the workplace.
- Credit and insurance: Systems that assess the creditworthiness of individuals, or price risk in life and health insurance, are high-risk. Deployers of these systems must also carry out a fundamental rights impact assessment before using them.
- Exported products: AI acting as a safety component in products such as machinery, lifts, toys, or medical devices counts as high-risk from 2 August 2028 and becomes part of the product's conformity assessment. Harmonised standards will make this easier: the reference of EN 18286:2026, the first standard on quality management systems for AI, was published in the Official Journal in 2026.
If you are a Turkish company that develops such a system and offers it to customers in the EU, your obligations are heavier: a risk management system, data governance, technical documentation, automatic logging, human oversight, accuracy and cybersecurity, a quality management system, conformity assessment, CE marking, registration in the EU database, and an authorised representative in the EU.
Related: LLM Observability and OpenTelemetry: Building Production Tracing for Enterprise AI Systems
Penalties
| Infringement | Maximum administrative fine |
|---|---|
| Prohibited practices | €35 million or 7% of worldwide annual turnover, whichever is higher |
| Other obligations, including high-risk and transparency | €15 million or 3% of turnover |
| Supplying incorrect, incomplete, or misleading information to authorities | €7.5 million or 1% of turnover |
For SMEs and start-ups, the lower of the two amounts applies. The omnibus extended this rule to small mid-cap companies for fines other than those for prohibited practices. Fines are set and enforced by the competent authorities of the EU member states; for providers of general-purpose AI models, the European Commission is responsible.
How the EU AI Act and KVKK work together
The AI Act does not replace data protection law. If an AI assistant processes personal data, three sets of rules can apply at once:
- KVKK applies to personal data processed in Türkiye. Sending personal data to an AI provider abroad is subject to KVKK's rules on transfers abroad.
- The GDPR can apply to you even if you are not established in the EU, when you offer goods or services to people in the EU or monitor their behaviour.
- The AI Act adds AI-specific obligations on top, such as transparency, risk management, and human oversight.
Türkiye's Personal Data Protection Authority has also published guidance: the "Generative AI and Protection of Personal Data Guide (in 15 Questions)" in November 2025, and a document on the use of generative AI tools in the workplace in March 2026. The latter notes that workplace use often follows individual preference rather than a defined company policy. A written AI acceptable use policy is the most concrete first step for both KVKK and the EU AI Act.
Related: AI Acceptable Use Policy: A Free Template for Companies
Where is Türkiye's own AI regulation?
As of October 2026, Türkiye has no AI law in force; bills on the subject are waiting at the committee stage in the Grand National Assembly. The Türkiye AI Action Plan (2026–2030), published in the Official Gazette in August 2026, provides for a National AI Board to monitor implementation, at least five regulatory sandboxes in priority sectors such as finance, health, energy, mobility, and telecommunications, and the preparation of a risk-based legal and ethical framework.
The practical conclusion for a Turkish company selling into the EU: waiting for Turkish rules to become clear is no reason to delay the EU obligations. An inventory, risk classification, and transparency practices built for the EU rules are also a solid starting point for whatever risk-based rules Türkiye adopts later.
Related: The Turkish AI & LLM Landscape in 2026: Who Is Building What
A 90-day compliance plan
Days 1–30: Visibility
- Build an inventory. List every AI system you use: website assistants, internal tools, AI features embedded in software, and models supplied by vendors.
- Map your EU exposure. For each system, note users, customers, subsidiaries, and exported products in the EU.
- Classify the risk level. Place each system in one category: prohibited, high-risk, transparency, or minimal risk. If you find a prohibited practice, stop it immediately.
Days 31–60: Urgent fixes
- Fix transparency. Complete chatbot notices, deepfake labels, and marking confirmations from vendors.
- Introduce a policy and training. Publish a written AI acceptable use policy, train teams in AI literacy, and keep training records.
- Review vendor contracts. Check data processing terms, model documentation, marking obligations, and, where needed, the authorised representative in the EU.
Days 61–90: High-risk preparation
- Draw up a roadmap for high-risk uses. If you use AI in HR, credit, or product safety, plan backwards from December 2027 and August 2028.
- Set up logging and oversight. Record AI decisions, define human oversight points, and write down the incident reporting process.
- Appoint an owner. Name the person responsible for AI compliance and have them track changes to the timeline.
Common mistakes
- "We are not in the EU, so it doesn't concern us." It does if your AI output is used in the EU.
- "The omnibus postponed everything." What moved was the high-risk timeline. The bans have applied since February 2025, and the transparency rules since August 2026.
- "A sentence in the privacy policy is enough." The information must be given clearly, at the latest at the first interaction.
- "Compliance is the software vendor's problem." Deployers have obligations too. For a system offered under your own brand, you may count as the provider and take on the provider's obligations.
- "If we comply with KVKK, we comply with the AI Act." The two ask for different things: one regulates personal data, the other the AI system itself.
Frequently asked questions
Does the EU AI Act apply to companies in Türkiye?
Yes, in certain cases. You are in scope if you place an AI system or model on the EU market, if your AI output is used in the EU, or if a company of yours established in the EU uses AI. If you use AI only for activities in Türkiye and do not use its output in the EU, the Act does not apply to you directly.
Does Türkiye have an AI law?
Not as of October 2026. Bills are pending in parliament, and the 2026–2030 Action Plan provides for a risk-based legal and ethical framework. For personal data, KVKK and the AI guidance of the Personal Data Protection Authority apply today.
What did the digital omnibus change?
Regulation (EU) 2026/1744 moved the high-risk deadline for Annex III systems to 2 December 2027 and for Annex I systems to 2 August 2028. It softened the AI literacy duty, extended some SME relief to small mid-cap companies, and added new bans covering non-consensual intimate deepfakes and child sexual abuse material. The transparency rules were not postponed.
Is our employees' use of ChatGPT covered by the Act?
If it is used only for internal work in Türkiye, it is usually out of scope. If a team in the EU uses it, or its output is used in the EU, tasks such as drafting and summarising are generally minimal risk, and the main expectation is measures that support staff AI literacy. If the same tool is used for a high-risk purpose, such as screening candidates in the EU, the high-risk rules apply. KVKK and data security rules apply in every case.
Do we need a representative in the EU?
Only if you are a provider placing a high-risk AI system or a general-purpose AI model on the EU market. For a website assistant that is only subject to the transparency rules, the AI Act sets no separate authorised representative requirement. The GDPR's own representative rules may still apply.
Are fines lower for SMEs?
Yes. For SMEs and start-ups, each fine is capped at the lower of the fixed amount and the percentage of turnover. The omnibus extended this rule to small mid-cap companies for fines other than those for prohibited practices.
Next steps
Complying with the EU AI Act is as much a technical job as a legal one: you need to know which system runs where, control what your assistants say, and keep a record of decisions. Orbitra helps you inventory your AI systems, identify the uses that touch the EU, and bring your website assistants in line with the transparency rules through AI automation consulting. We build your teams' AI literacy through corporate AI training.
This article is general information and not legal advice. Consult a lawyer about your specific situation.
