EU AI Act and AI Governance: Timeline After the Digital Omnibus and a Practical Checklist
AI governance is the set of policies, roles, processes and controls that make an organisation's use of AI lawful, safe and accountable. In Europe, its legal backbone is the EU AI Act, the first comprehensive AI regulation, whose timeline was amended in 2026 by the Digital Omnibus on AI.
Key takeaways
- The AI Act classifies AI by risk: prohibited, high-risk, transparency obligations, and minimal risk; general-purpose AI models have their own rules.
- The Digital Omnibus, in force since 27 July 2026, postponed high-risk obligations to 2 December 2027 (Annex III) and 2 August 2028 (Annex I).
- Prohibitions, AI literacy and general-purpose AI obligations already apply; transparency obligations start in 2026.
- A governance programme built on an AI inventory, risk classification and controls aligned with NIST AI RMF and ISO/IEC 42001 prepares you for every deadline.
The risk-based structure
| Category | Examples | Consequence |
|---|---|---|
| Prohibited practices | Social scoring, manipulative techniques, certain biometric uses; the Omnibus added non-consensual intimate imagery and child sexual abuse material generators | Banned in the EU. |
| High-risk AI | AI in recruitment, credit scoring, education, essential services, critical infrastructure (Annex III), or as a safety component of regulated products (Annex I) | Risk management, data governance, documentation, logging, human oversight, accuracy, robustness and cybersecurity, conformity assessment. |
| Transparency obligations | Chatbots, systems generating synthetic content, deepfakes | Inform users they interact with AI; mark synthetic content in a machine-readable way. |
| Minimal risk | Spam filters, most productivity uses | No specific obligations beyond general law and AI literacy. |
| General-purpose AI models | Foundation models and LLMs offered to others | Technical documentation, copyright policy, training-data summary; extra duties for models with systemic risk. |
Timeline (updated by the Digital Omnibus)
| Date | What applies |
|---|---|
| 1 August 2024 | The AI Act enters into force. |
| 2 February 2025 | Prohibited practices and AI literacy obligations apply. |
| 2 August 2025 | Obligations for general-purpose AI models apply; governance and penalty provisions start. |
| 27 July 2026 | The Digital Omnibus on AI enters into force and amends the timeline. |
| 2 August 2026 | Transparency: users must be told they are interacting with an AI system; synthetic-content marking for systems placed on the market from this date. |
| 2 December 2026 | Synthetic-content marking for systems placed on the market before 2 August 2026. |
| 2 August 2027 | Member States must have at least one AI regulatory sandbox. |
| 2 December 2027 | High-risk obligations for stand-alone systems (Annex III). |
| 2 August 2028 | High-risk obligations for AI embedded in products covered by EU sectoral safety law (Annex I). |
This summary is for information only and is not legal advice. Check the consolidated text in the Official Journal of the EU for your specific case.
Other changes in the Digital Omnibus
- AI literacy: providers and deployers must take measures to support AI literacy of relevant staff, without a guaranteed individual level; deployers of high-risk systems keep specific competence requirements.
- Registration: the EU database obligation for systems assessed under Article 6(3) remains, with simplified information requirements; the assessment must be documented before market placement.
- SMEs and small mid-caps: regulatory support, including priority access to sandboxes, now extends to small mid-cap companies.
Building an AI governance programme
Two voluntary frameworks give a structure that also prepares for the AI Act:
- NIST AI Risk Management Framework — four functions: Govern, Map, Measure, Manage.
- ISO/IEC 42001 — a certifiable AI management system standard, in the same family as ISO/IEC 27001.
A practical checklist
- Inventory every AI system, model, agent and AI-enabled SaaS feature in use, with an owner.
- Classify each one under the AI Act (prohibited, high-risk, transparency, minimal) and record your role (provider or deployer).
- Assign accountability: an AI governance owner, a review board and clear escalation paths.
- Train staff to meet AI literacy expectations.
- Assess risks for each system — security, privacy, bias, reliability — and test them, including through red teaming.
- Implement controls: human oversight, logging, access control, data governance, runtime guardrails.
- Document decisions, tests and controls so you can demonstrate compliance.
- Monitor systems in production and handle incidents.
- Review regularly, and whenever a model, use case or regulation changes.
Frequently asked questions
When do EU AI Act high-risk obligations apply?
After the Digital Omnibus on AI, high-risk obligations apply from 2 December 2027 for stand-alone systems listed in Annex III, and from 2 August 2028 for AI embedded in products covered by EU sectoral safety legislation (Annex I).
What is the Digital Omnibus on AI?
An amending regulation that entered into force on 27 July 2026. It postponed high-risk deadlines, clarified AI literacy, simplified some registration requirements, extended support to small mid-caps and added new prohibited practices.
Does the AI Act apply to companies outside the EU?
Yes, when they place AI systems on the EU market or when the output of their AI systems is used in the EU.
What is the difference between NIST AI RMF and ISO/IEC 42001?
NIST AI RMF is a voluntary US risk-management framework organised around Govern, Map, Measure and Manage. ISO/IEC 42001 is an international, certifiable management-system standard for AI. Many organisations use NIST AI RMF for risk practices and ISO/IEC 42001 for certification.
Where should AI governance start?
With an inventory of AI systems and their owners. Classification, risk assessment and controls all depend on knowing what AI is in use.
Related guides
- AI Agent & MCP Security — Threats to agents and MCP servers, with a practical checklist.
- AI Observability — What to monitor in LLM apps and agents, with OpenTelemetry.
- OWASP Top 10 for LLM — The ten critical LLM risks, controls, and what changed in 2026.