The agreement has been politically finalized and published but requires formal adoption before taking effect. The original timeline called for high-risk AI obligations to begin in August 2026; the new framework extends that window by 16 to 24 months depending on the use category.
The delay affects rules governing high-impact applications including employment screening, biometric identification, critical infrastructure management, education, and migration control. Organizations relying on AI in these areas should monitor the formal adoption process and use the extended timeline to audit current systems against the Act's requirements. The compliance window remains compressed for firms not yet prepared; the delays do not eliminate obligations, only postpone them.