It entered into force in August 2024 with obligations phasing in through 2027. Transparency and copyright duties for general-purpose AI models — including extra requirements for 'systemic risk' models above a compute threshold — took effect from August 2025; high-risk system obligations phase in later. It applies to any provider serving users in the EU regardless of where the company is headquartered, making it a de facto global compliance baseline alongside the US's state-level rules like California's SB 53.