EU AI Act Enters Into Force: Timeline and Implementation Phases

A Historic Day for AI Regulation

On August 1, 2024, the European Union's Artificial Intelligence Act officially entered into force, marking a pivotal moment in the global regulation of AI technology. This is the world's first comprehensive AI law, setting the stage for how AI systems will be developed, deployed, and governed in the EU.

What "Entering Into Force" Means

It's crucial to understand that entering into force is NOT the same as full application. The Act follows a phased implementation approach:

Entry into force (August 1, 2024): The law is officially part of EU legislation
Application dates (various): Different provisions become enforceable at different times

The Implementation Timeline

Phase 1: Prohibitions (February 2, 2025)

6 months after entry into force

Ban on prohibited AI practices takes effect
Includes social scoring, exploitation of vulnerabilities, subliminal manipulation
No grace period - systems must be decommissioned

Phase 2: General-Purpose AI (August 2, 2025)

12 months after entry into force

Obligations for general-purpose AI models begin
Foundation models must comply with transparency requirements
Systemic risk models (>10^25 FLOPs) face additional obligations

Phase 3: Annex I Systems (August 2, 2026)

24 months after entry into force

Existing high-risk systems listed in Annex I must comply
Full conformity assessment requirements
CE marking mandatory

Phase 4: Full Application (August 2, 2027)

36 months after entry into force

All provisions fully applicable
Complete enforcement of all high-risk system requirements
Member state penalties in effect

Immediate Actions Required

For All Organizations

Audit your AI systems - Identify what you're using/developing
Classify risk levels - Use our classifier tool to assess each system
Review prohibited practices - Ensure none of your systems fall into banned categories
Create implementation roadmap - Plan for compliance based on deadlines

For High-Risk System Providers

Begin documentation processes now
Establish quality management systems
Plan for conformity assessments
Budget for compliance costs

For GPAI Developers

Prepare technical documentation
Implement content policies
Plan for model evaluations
Consider open-source implications

Key Provisions Now in Legal Effect

While not yet enforceable, these provisions are now law:

Definitions and Scope

AI system definition established
Risk categories officially defined
Territorial scope clarified (extraterritorial application)

Institutional Framework

AI Board establishment process begins
National competent authorities designation starts
Standards development initiated

Market Surveillance

Framework for enforcement established
Penalty structures defined (up to €35M or 7% global turnover)
Complaint mechanisms outlined

What's NOT Yet Required

Important: The following are NOT yet mandatory:

Technical documentation (except for new systems)
Conformity assessments
CE marking
Registration in EU database
Most transparency obligations

Preparing for What's Next

Q3-Q4 2024

Finalize AI inventory
Complete risk assessments
Begin documentation templates
Establish governance structures

Q1 2025

Prepare for prohibition deadline (February 2)
Start GPAI compliance preparations
Initiate staff training programs

Throughout 2025

Implement technical requirements
Conduct gap analyses
Engage with conformity assessment bodies
Monitor regulatory guidance

Resources and Support

The European Commission will provide:

Harmonized standards (ongoing development)
Guidelines and best practices
Regulatory sandboxes
Support for SMEs and startups

The Global Impact

The EU AI Act's entry into force signals:

Brussels Effect: Global companies will likely adopt EU standards worldwide
Regulatory inspiration: Other jurisdictions following EU's lead
Market advantage: Early compliance can become competitive differentiator
Innovation framework: Clear rules enabling responsible AI development

Conclusion

August 1, 2024, marks the beginning of a new era in AI governance. While full compliance isn't required immediately, the clock is now ticking. Organizations have a unique opportunity to get ahead of requirements and establish themselves as leaders in responsible AI.

The phased approach provides time to adapt, but don't be complacent. The February 2025 prohibition deadline will arrive quickly, and high-risk system requirements in 2026 require substantial preparation.

Start your compliance journey today with our risk assessment tool to understand exactly how the EU AI Act applies to your AI systems.