Friday marked the activation of the European Union AI Act's obligations for high-risk AI systems, the deadline that turns the law from principle into compliance work. Providers of systems used in employment screening, credit decisioning, education access, biometric identification and critical infrastructure must now maintain conformity assessments, technical documentation and post-market monitoring plans.

The European Commission's AI Office confirmed that registration in the public EU database is open and that the first batch of national notified bodies has been designated across Germany, France, Ireland, the Netherlands and Spain. Penalties for non-compliance scale up to 15 million euro or 3 percent of global turnover, whichever is higher.

Industry response has been uneven. Larger vendors had documentation ready, while several mid-market SaaS providers have temporarily geofenced affected features from EU customers while paperwork is completed. The Commission has indicated it will prioritise enforcement against systemic risk and clear bad faith over technical filing errors in the first twelve months.