AI Act in 2026: what companies should already have in place and what comes next

The EU regulation on artificial intelligence (the AI Act, Regulation 2024/1689) applies in stages. It changed in July 2026: the omnibus package Regulation (EU) 2026/1744, published on 24 July 2026, postponed the deadlines for high-risk systems and amended the AI literacy duty.

Sources: Regulation (EU) 2024/1689, Artificial Intelligence Act (in Czech), Regulation (EU) 2026/1744, digital omnibus on AI (in Czech)

What has applied since 2025?

Since 2 February 2025 the general provisions have applied, including the definitions, the AI literacy duty (Article 4) and the prohibited practices (Article 5). Prohibited practices include, for example, social scoring and the untargeted scraping of facial images from the internet or CCTV footage to build facial recognition databases.

Since 2 August 2025 the rules for providers of general-purpose AI models (large language models and similar) have applied, and Member States were required to designate supervisory authorities and set rules on penalties.

Sources: Regulation (EU) 2024/1689, Artificial Intelligence Act (in Czech), European Commission, AI Act Service Desk: implementation timeline

What changed on 2 August 2026?

From 2 August 2026 the regulation applies generally. For most companies two things matter:

• Transparency (Article 50). People must know that they are interacting with an AI system, for example a chatbot on a website. Artificially generated content is marked in a machine-readable way. Anyone publishing a deepfake must disclose it.
• Enforcement. According to the European Commission, enforcement has begun for the prohibitions, transparency, AI literacy and general-purpose models.

So if you use a chatbot for customers or generate content with AI, these rules may already apply to you. Some obligations fall on the provider of the system, some on whoever uses it. It depends on your role.

Sources: Regulation (EU) 2024/1689, Artificial Intelligence Act (in Czech), European Commission, AI Act Service Desk: implementation timeline

What did the 2026/1744 package change?

The package, known as the digital omnibus on AI, made three changes that matter for companies:

1. Later deadlines for high-risk systems. Rules for systems under Annex III (for example AI in employment and recruitment, education, critical infrastructure or access to essential services) apply from 2 December 2027. Rules for AI as part of regulated products under Annex I apply from 2 August 2028. They were originally due to apply from 2 August 2026 and 2 August 2027 respectively.
2. An amended AI literacy duty. Originally, providers and deployers had to ensure, to their best extent, a sufficient level of AI literacy. Article 4 now requires them to take measures to support the improvement of AI literacy among their staff and other persons dealing with the operation and use of AI systems on their behalf. It expressly does not require a specific level of literacy to be guaranteed for individuals. The Commission and Member States are to support companies in meeting the duty, especially small and medium-sized ones, and the Commission will publish practical examples. We discuss the duty in detail in the article AI literacy for employees.
3. New prohibitions from 2 December 2026. Systems whose purpose or function is to create or alter realistic intimate or sexually explicit depictions of an identifiable person without their explicit consent will be prohibited, as will systems that generate child sexual abuse material.

Source: Regulation (EU) 2026/1744, digital omnibus on AI (in Czech)

What comes next for companies?

The later deadlines do not mean you can wait until 2027 to start. High-risk systems require risk management, high-quality data, records, technical documentation and human oversight, and preparing all of this takes time.

Sources: Regulation (EU) 2026/1744, digital omnibus on AI (in Czech), Regulation (EU) 2024/1689, Artificial Intelligence Act (in Czech), European Commission, AI Act Service Desk: implementation timeline

Who will enforce the AI Act in Czechia?

As far as available sources show, the Czech Artificial Intelligence Act had not been adopted as of 31 October 2026. The Government Legislative Council discussed the draft on 6 August 2026 and adjourned the discussion. In a press release of 31 July 2026, the Czech Telecommunication Office (ČTÚ) states that the transparency rules will be supervised by market surveillance authorities, in Czechia including ČTÚ, and in some cases by the European AI Office. The detailed division of supervision between authorities will only be set by the act. We will update this part of the article.

Sources: Government Legislative Council, draft Artificial Intelligence Act (in Czech), Czech Telecommunication Office, transparency rules under the AI Act (in Czech)

What fines apply?

Under Article 99 of the regulation, prohibited practices carry a fine of up to EUR 35 million or 7% of worldwide annual turnover, and breaches of most other obligations, including transparency, up to EUR 15 million or 3% of turnover. The higher of the two amounts applies. For small and medium-sized enterprises, including start-ups, the lower of the two applies instead. Regulation 2026/1744 extended this rule to small mid-cap companies, though not for prohibited practices.

Sources: Regulation (EU) 2024/1689, Artificial Intelligence Act (in Czech), Regulation (EU) 2026/1744, digital omnibus on AI (in Czech)

What can a company check now?

We recommend starting with three steps:

1. An inventory of AI in the company. Which AI systems you use or develop, what for, and whether you supply them to others. This can include AI features inside everyday software.
2. Role and risk. For each system, determine whether you are the provider or the deployer and which risk category the use falls into. Pay particular attention to recruitment and employee assessment, which Annex III lists among high-risk uses.
3. Transparency and literacy. Check that your chatbot or generated content meets the rules in Article 50, and have a simple plan for ongoing training of the people who work with AI.

Summary

• Since 2 August 2026 the transparency rules have applied and, according to the European Commission, enforcement has begun. Chatbots and AI-generated content may already be relevant, depending on the company's role.
• Regulation 2026/1744 postponed high-risk systems to 2 December 2027 (Annex III) and 2 August 2028 (Annex I) and softened the AI literacy duty.
• The Czech AI act has not yet been adopted. Transparency will be supervised by market surveillance authorities, in Czechia including ČTÚ.

An overview of deadlines broken down by role is on the page AI Act: your company's readiness. If you want to check which obligations apply to you, or take an AI project into production in line with the rules, see our AI solutions or write to us.

Status as of 31 October 2026. This article is general information, not legal advice. We update it monthly based on the European Commission and the Official Journal of the EU.

• AI Act
• AI regulation
• AI literacy