EU delays AI duties, but transparency still comes soon

What this is about

The European Commission updated its central AI Act page on July 24, 2026. The key point: the AI Omnibus package is now in force, delaying parts of the high-risk regime while keeping pressure on transparency obligations.

For companies in Germany and across Europe, that creates an awkward mix. Some large compliance projects get more time. Other duties, such as notices for chatbots, deepfakes and AI-generated content, remain close on the calendar.

What the AI Act timeline actually does

The AI Act applies in stages. Bans and AI literacy duties have applied since February 2025. Rules for general-purpose AI models have applied since August 2025. Transparency rules are set to apply in August 2026, while certain marking and detection duties for systems already on the market get a limited grace period until December 2, 2026.

The Omnibus package mainly moves high-risk systems. Stand-alone systems in areas such as biometrics, critical infrastructure, education, employment, migration and border control are set to apply from December 2, 2027. Systems embedded in products, such as machinery or toys, follow on August 2, 2028.

Why it matters

This is not only a legal-department issue. Any company using a customer-service chatbot, an AI tool in HR or automatically generated text for public information must check whether users can clearly see when they are interacting with AI or viewing AI content.

At the same time, Europe gets more time to put technical standards, sandboxes and supervision into a workable shape. That makes sense because high-risk compliance without harmonised standards can quickly become expensive and unclear. The downside is that companies may wrongly read the delay as a full pause.

In plain language

Imagine an airport renovation. New security checks for special areas open later because the devices and rules still need coordination. But signs at the entrance still need to be up on time so travelers know where they are and who they are speaking to.

A practical example

A German mid-sized company uses a chatbot for 30,000 customer requests per month from September 2026 and an internal tool that pre-sorts job applications. The chatbot needs clear notices that users are speaking with AI. The hiring tool may later need stricter documentation, monitoring and human oversight depending on its exact use. Both belong in the same implementation plan, but not under the same deadline pressure.

Scope and limits

• The specific duties depend heavily on whether a company is a provider, deployer or user of a system.
• The delay does not affect every rule; transparency remains a near-term issue.
• National supervision, technical standards and practical audit processes are still evolving, so companies should not treat a one-off review as enough.

The sober conclusion: the AI Act becomes less abrupt through the Omnibus package, but not harmless. Anyone looking only at high-risk deadlines before August 2026 may miss the visible user-facing duties.

SEO & GEO keywords

EU AI Act, AI Omnibus, Article 50, transparency obligations, high-risk AI, AI compliance, European Commission, AI Office, deepfake labelling, chatbot notice, Germany, European Union