Compliance 5 min read

EU AI Act Compliance Timeline: Key Dates You Can't Miss

A complete breakdown of the phased implementation of the AI Act. From prohibited practices to general-purpose AI models, know exactly when your obligations kick in.

By Elena Rossi, Legal Tech Analyst · Published

The Clock is Ticking

The EU AI Act has officially entered into force, triggering a series of compliance deadlines that organizations must adhere to. Missing these dates can result in significant penalties, not to mention reputational damage. This guide breaks down the critical milestones.

6 Months: Prohibited AI Practices

By February 2, 2025, the ban on "unacceptable risk" AI systems takes effect. This includes social scoring systems, biometric categorization using sensitive characteristics, and untargeted scraping of facial images for databases.

12 Months: General-Purpose AI (GPAI)

Providers of General-Purpose AI models (like large language models) face a deadline of August 2, 2025. They must maintain technical documentation, comply with EU copyright law, and for those with "systemic risk," conduct adversarial testing and incident reporting.

24 Months: High-Risk AI Systems

Most obligations for high-risk AI systems (Annex III) apply by August 2, 2026. This is the big one for most enterprises. Requirements include:

  • Implementing a Risk Management System
  • Data Governance frameworks
  • Technical Documentation
  • Human Oversight measures
  • Accuracy, Robustness, and Cybersecurity

36 Months: Product Safety Component AI

For AI systems that are safety components of products already regulated under EU law (like medical devices, cars, machinery), the deadline is August 2, 2027, aligning with relevant product safety legislation.

Start Now

Waiting until the deadline is a strategy for failure. Gap analysis and inventory mapping should be happening today.