EU AI Act delayed: what the 2026 changes mean for your business

For months, companies that sell into Europe had one date circled on the calendar: August 2, 2026, when the strictest parts of the EU's Artificial Intelligence Act were due to start. That date has now moved. In 2026 the EU adopted a package of changes, known as the "Digital Omnibus on AI", that delays the rules for high-risk AI systems and simplifies several other obligations.

If you run a factory or a business in Turkey that exports to the EU, or that uses AI tools in products and services for European customers, this is worth a few minutes of your time. Below we explain in plain terms what changed, which dates now matter and what a sensible next step looks like.

What happened, in short

The European Commission proposed the simplification package on November 19, 2025. Parliament and Council reached a political agreement on May 7, 2026, according to the Commission. The European Parliament adopted its position on June 16, 2026, by 423 votes to 57 with 174 abstentions, and the Council gave its final green light on June 29, 2026.

The result is Regulation (EU) 2026/1744, published in the Official Journal of the EU. According to the Commission, it entered into force on July 27, 2026. It amends the original AI Act rather than replacing it, so the law itself still stands.

The new dates that matter

The official reason for the delay is practical. The regulation states that standards, common specifications and guidance were not ready in time, and that national authorities had not been set up, which risked a sharp rise in compliance costs. The new timeline is:

• December 2, 2027: Rules start for stand-alone high-risk AI systems. These are the use cases listed in the law, such as biometrics, critical infrastructure, education, employment, migration and border control.
• August 2, 2028: Rules start for high-risk AI built into products that are already covered by EU safety laws, such as lifts or toys. For machinery and equipment makers, this is the date to watch.
• December 2, 2026: The Council says the grace period for providers to add transparency solutions for AI-generated content was cut from six months to three, with this new deadline.
• Already in force: Bans on certain AI practices and the AI literacy duty have applied since February 2, 2025, and the rules for general-purpose AI models since August 2, 2025, according to the Commission.

The package also added a new ban on AI practices that generate non-consensual intimate content or child sexual abuse material.

What else changed for smaller companies

The changes are not only about dates. Several points directly affect small and mid-sized firms.

AI literacy is softer, but still a duty. The original text required companies to ensure a "sufficient level" of AI literacy among their staff. The amended article now asks providers and deployers to take measures that support AI literacy, taking into account the knowledge, experience and context of the people involved. The Commission's question-and-answer page confirms that it remains an obligation, but no specific level is mandated. The Commission and member states are also expected to support companies, especially smaller ones, with practical examples and training.

Simpler paperwork for growing companies. The regulation introduces a definition for "small mid-cap enterprises", companies that have outgrown the SME category, and extends some SME relief to them. According to the Parliament summary, SMEs, start-ups and small mid-caps may use simplified technical documentation for high-risk AI systems.

A narrower idea of "high risk" for products. The Parliament summary notes that the definition now targets AI systems likely to have a negative impact on the health and safety of people or property. Products with AI functions that only assist users or optimize performance will not automatically face high-risk obligations, as long as their failure does not create health or safety risks.

Why this matters for businesses in Turkey

Turkey is not an EU member, but the AI Act follows the product and its use, not the company's address. If you place an AI system on the EU market, or the output of your AI system is used in the EU, the rules can apply to you. That is the same logic many exporters already know from CE marking.

In practice, three groups should pay attention:

• Machinery and equipment makers who add AI features, such as vision-based quality checks or predictive maintenance, to products sold in Europe.
• Software and service providers whose tools are used by European customers.
• Suppliers to European manufacturers, who may be asked by their customers for documentation about how their AI works.

What to do now: a practical checklist

A delay is not a cancellation. It is extra time to prepare calmly instead of in a rush.

1. List your AI use. Write down every place where your company uses or sells AI: chat assistants, quality inspection cameras, demand forecasting, recruitment tools. Most will be low risk, and that is good to know.
2. Mark what touches the EU. For each item, note whether it is sold to, or used for, customers in the EU.
3. Check for high-risk areas. Compare your list with the high-risk areas named by the Commission and with the EU safety laws that already cover your products. If you are unsure, ask your notified body or a legal advisor.
4. Start a light AI literacy plan. A short internal guide, a training session and clear rules on which tools staff may use are reasonable first measures. They are useful even without any legal duty.
5. Keep simple records. Note which AI tools you use, for what purpose and who is responsible. Good records make any later compliance work faster and cheaper.
6. Watch for standards and guidance. The new dates depend on guidance and standards arriving. Follow the Commission's AI Act page for updates.

Frequently asked questions

Has the AI Act been cancelled? No. Only some dates moved and some obligations were simplified. Bans on certain practices and the AI literacy duty already apply.

Does my small business need to hire a compliance team? Not necessarily. Most everyday business uses of AI are not high risk. A clear inventory and basic staff training are a sensible start for most companies.

Is this legal advice? No. This post summarizes official EU sources. For a decision about a specific product, consult a qualified advisor.

Conclusion

The EU has given companies more time, especially for high-risk AI built into products, now due on August 2, 2028. For exporters in Turkey, the smart move is to use this time to map AI use, train teams and keep records, so the deadline does not turn into a rush.

If you want help building a simple inventory of where AI and automation are used in your business, and which parts may touch EU rules, AtomAxis can help you get started.