Most provisions of the EU AI Act have applied since 2 August 2026. Further deadlines follow in 2027, most importantly 2 December 2027, when obligations for high-risk AI under Annex III take effect. This article explains what already applies, what follows and how to plan the year. Information as of 15 November 2026. Source: European Commission timeline.
What already applies
Prohibitions on certain AI practices and the AI literacy rule in Article 4 have applied since 2 February 2025. Article 4 has applied in its amended form since 27 July 2026: providers and deployers take measures supporting their staff's AI literacy.
Obligations for general-purpose AI models have applied since 2 August 2025. Most provisions of the regulation, including the transparency obligations in Article 50, have applied since 2 August 2026.
On 2 December 2026, the transition period for labelling generated content in older systems ends and two new prohibitions take effect. See AI labelling requirements from 2 December 2026 for details.
Deadlines in 2027 and 2028
2 August 2027: each member state should have at least one AI regulatory sandbox in operation, where companies can develop and test AI systems under authority supervision.
2 December 2027: obligations for high-risk AI under Annex III apply. The original date was 2 August 2026; Regulation (EU) 2026/1744 postponed it.
2 August 2028: obligations for high-risk AI in products under Annex I apply, for example AI used as a safety component in machinery or medical devices.
Which uses fall under Annex III
Annex III lists eight areas: biometrics, critical infrastructure, education and vocational training, employment and workforce management, access to essential private and public services, law enforcement, migration and border control, and administration of justice and democratic processes.
Two areas are particularly relevant to SMEs. First, HR: AI targeting job advertisements, screening applications or evaluating candidates, and systems involved in promotion, dismissal or task allocation decisions or monitoring performance and behaviour.
Second, finance and insurance: AI assessing creditworthiness or establishing credit scores, except for fraud detection, and systems assessing risks and prices in life and health insurance.
Whether a system qualifies as high-risk AI depends on its specific intended purpose. A language model used for everyday office work is not automatically high-risk. Using the same model to select job applicants may change that assessment.
What deployers of high-risk AI must do from December 2027
Article 26 sets out the duties of deployers: companies using a high-risk system. These include following the provider's instructions, assigning human oversight to people with the necessary competence, training and authority, and monitoring operation.
Deployers must retain automatically generated logs under their control for at least six months, unless other rules specify otherwise. Employers must inform workers' representatives and affected employees before using a high-risk system in the workplace.
Your roadmap for 2027
January to March: inventory all AI systems and clarify whether you are a provider or deployer. Check whether systems operate in HR, finance or other Annex III areas.
April to June: assign responsibilities, update your AI policy and document AI literacy measures. Ask providers of high-risk systems for instructions and evidence of conformity.
July to September: plan human oversight, logging and information for the works council and employees. Include these requirements from the outset in new HR projects.
October to 2 December 2027: resolve outstanding items and check that all high-risk uses meet deployer obligations.
Who is responsible in Germany
Under section 2 KI-MIG, the Federal Network Agency is generally the competent market surveillance authority. Specific responsibilities remain, including for regulated financial activities and public bodies of the German states. Source: KI-MIG, Federal Law Gazette 2026 I No. 223.
Frequently asked questions about the EU AI Act in 2027
Were high-risk AI deadlines postponed? Yes. Under Regulation (EU) 2026/1744, obligations for high-risk AI under Annex III apply from 2 December 2027, and under Annex I from 2 August 2028.
Does the AI Act also apply to small companies? Yes, regardless of company size. The obligations depend on your role and the system's intended purpose.
Is ChatGPT in a company high-risk AI? Not simply because it is used in a company. The application matters. Selecting applicants with AI, for example, falls under Annex III.
What should I do first in 2027? Create a complete list of your AI systems with roles and intended purposes. Without it, obligations cannot be reliably assigned.
Support with planning
In an AI compliance assessment, we work with you to inventory AI systems, classify roles and risk areas and plan the steps through December 2027. The assessment costs from €4,990 plus VAT. Case-specific legal review remains with your legal adviser. Read about EU AI Act consulting or book an initial consultation.
Sources
European Commission AI Act Service Desk: implementation timeline, Article 26, Annex III, Regulation (EU) 2026/1744 and KI-MIG. Accessed on 15 November 2026. This article does not replace legal advice.