Blog · Türkiye · Regulation
Türkiye's AI Action Plan 2026-2030: what Turkish companies should do now
21 August 2026 · 4 min read · By LeadAI
In August, Türkiye's Artificial Intelligence Action Plan for 2026-2030 moved into its implementation phase. It is the clearest sign yet of where Turkish AI rules are heading. The short answer for companies is this: the plan does not create direct legal duties for you today, but it tells you what the duties are likely to look like, so it is worth preparing now.
What was published
Presidential Circular No. 2026/9, dated 17 August 2026, was published in the Official Gazette (No. 33344) on 18 August 2026. The plan had earlier been presented in June. According to law-firm summaries, it contains 16 priority actions under four pillars: Recognise, Leverage, Produce and Govern. It also sets out targets for skills, infrastructure and investment, and calls for a multi-stakeholder National AI Ethics Board and regulatory sandboxes in priority sectors. It does not name a single stand-alone AI regulator.
What it says about rules
The part that matters most for companies is Action 4, a planned national legal and ethical framework for AI. As described by law firms that have analysed it:
- It takes a risk-based approach in the spirit of the EU's, with obligations that grow with the level of risk.
- A short-form algorithmic impact assessment is envisaged for medium-risk applications and a longer one for high-impact systems.
- High-impact systems would need standardised technical documentation and a publicly available summary model card.
Nothing here is binding on private companies yet. The detail will arrive through secondary legislation and sector-specific guidance, which is where the real compliance agenda will be set.
What already applies to you
Türkiye has no stand-alone AI law today. A 2024 draft AI bill (No. 2/2234) was still a draft in the public sources we reviewed, and draft amendments in late 2025 and early 2026 would extend platform responsibility for AI-generated content under the internet law (No. 5651) and the personal data law (KVKK, No. 6698). In practice, the main law that reaches AI today is KVKK, which governs AI whenever personal data is processed and gives people a right to challenge results based solely on automated processing. The KVKK has also published guidance on generative AI and personal data.
Five things to do now
1. Extend your data inventory to cover AI. Review your KVKK data processing inventory so it includes AI use cases, including tools staff use with personal accounts. Our guide to building an AI inventory shows what to record.
2. Spot likely high-impact systems. Mark systems that influence people's access to jobs, credit, services, safety or rights. These are the ones most likely to attract impact assessments and documentation duties.
3. Practise the paperwork now. Write a short impact assessment and a plain-language summary for your most important system. ISO/IEC 42001 already asks for AI system impact assessments, so this builds habits you will need anyway.
4. Name owners and approval routes. For each system, record who owns it, who approves changes and who can stop it. This answers most questions regulators and customers ask first.
5. Watch for secondary legislation. Assign one person to track implementing rules, sector guidance and the Ethics Board. Check the official text of the Circular rather than relying on summaries alone, because one legal review noted that the full plan was not yet available as a standalone document when it was written.
How it compares with the EU
| Türkiye | EU | |
|---|---|---|
| Status | Policy plan; binding rules still to come | Binding law, with dates now fixed |
| Approach | Risk-based tiers planned | Risk-based tiers in force |
| High-risk dates | Not yet set | 2 December 2027 and 2 August 2028 |
| Documentation | Impact assessments and public model cards planned | Technical documentation and conformity duties |
If you sell into the EU, the EU rules apply to you regardless. See what changed in the EU AI Act delay. If you operate only in Türkiye, EU practice is still the best guide to what is coming, because the plan explicitly takes the EU's proportionate approach into account.
Where to begin
Start with a clear picture of what you use. The free AI Readiness Check takes about 25 minutes, and the LeadAI Assure journey shows how an inventory grows into the evidence an ISO/IEC 42001 management system needs.
This article gives general information and is not legal advice or a substitute for ISO/IEC 42001:2023, Turkish legislation or professional advice. Check the official text of the Circular and current KVKK guidance.
Sources
- COTT, Türkiye's National AI Action Plan 2026-2030: https://cottgroup.com/en/legislation/item/turkiyes-national-artificial-intelligence-action-plan-2026-2030-a-new-era-for-the-private-sector
- Moroğlu Arseven, The Presidential Circular on the Action Plan: https://www.morogluarseven.com/insights/publications/the-presidential-circular-on-the-turkiye-artificial-intelligence-action-plan-was-published/
- CMS, Türkiye unveils new AI Action Plan: https://cms.law/en/tur/legal-updates/turkiye-unveils-new-ai-action-plan
- CMS, draft KVKK amendment on AI-generated content: https://cms.law/en/tur/legal-updates/tuerkiye-proposes-amendment-to-data-protection-law-introducing-platform-liability-for-ai-generated-content
- Legal 500, AI regulation in Türkiye: https://www.legal500.com/guides/?p=145418