EU AI Act Postponed, but Waiting Is Not a Strategy: What Companies Should Do Now
This summer brought a major change to the EU AI Act: the Digital Omnibus on AI entered into force on 27 July 2026 and postponed the start date for high-risk AI obligations. Many companies read that as breathing room. But look at the details and you’ll see that significant parts of the Act were not postponed, and some obligations already apply today. This article summarizes what changed, what didn’t, and what companies should focus on now.
What Was Postponed?
- Stand-alone high-risk systems (Annex III): For systems in areas such as biometrics, employment, credit scoring, education, and critical infrastructure, obligations now start on 2 December 2027 instead of 2 August 2026.
- High-risk AI embedded in regulated products (Annex I): For systems in products such as medical devices, machinery, and vehicles, the date moved to 2 August 2028.
The main reason for the delay is that the harmonized technical standards needed for compliance weren’t ready in time. In other words, the weight of the obligations didn’t decrease; only the calendar moved.
What Didn’t Change?
- Transparency obligations (Article 50): Duties such as telling users they’re interacting with an AI system and labeling AI-generated content have applied since 2 August 2026. Only the marking of synthetic content for systems already on the market before that date received extra time, until 2 December 2026.
- General-purpose AI (GPAI) models: Obligations for these models have applied since August 2025.
- Prohibited practices: The AI uses banned under the Act continue to apply on the same schedule.
In short: a company that uses AI for chatbots, content generation, or customer communication already faces the transparency rules, regardless of the high-risk postponement.
Why It Matters for Companies in Turkey
The AI Act can also apply to providers and users who place AI systems on the EU market or whose output is used in the EU; where a company is established isn’t the only factor. Turkish firms that sell products and services in Europe, especially manufacturers adding AI components to their products, should keep a close eye on this timeline.
5 Steps You Can Take Today
1. Build an AI inventory. Which AI systems are in use across the company, and which are embedded in your products and services? Purchased tools, in-house builds, and services teams adopted on their own all count. Visibility is the first step for everything else.
2. Classify risk. Whether a system is high-risk is determined by the Act’s risk-based classification, regardless of the postponement. Uses that touch areas such as employment, credit, or education deserve particular care.
3. Apply transparency rules now. Inform users in chatbots, and plan your marking infrastructure for AI-generated text, images, and audio with dates like 2 December 2026 in mind.
4. Lay the governance and documentation groundwork. Risk management, technical documentation, human oversight, logging, and monitoring take a long time to prepare. The postponement gives extra time to start, not a reason to hold off.
5. Clarify supplier and model-provider relationships. If you use third-party models and tools, spell out in contracts who is responsible for what. Using a system for a purpose the provider didn’t intend can shift responsibilities.
Conclusion: The Delay Is a Preparation Window
The Digital Omnibus extended the timeline for high-risk systems, but it didn’t change the Act’s structure, its risk-based approach, or its core obligations. Work like inventory, classification, and governance takes time, so the most effective approach is to use the extra months for steady, unhurried preparation. Companies that do will gain not just compliance, but trust and scalability in their enterprise AI projects.
This article is for general information only and is not legal advice. We recommend seeking qualified legal counsel on the scope of your systems and your specific obligations.
Our related service: https://www.ranna.com.tr/en/software-development-service/