
EU formally delays high‑risk AI compliance deadline but keeps strict data governance requirements
The EU has adopted Regulation 2026/1744, postponing the AI Act’s high‑risk system obligations that were due to apply on August 2, 2026, after delays in standards and national supervisory capacity. The law pushes key requirements to a later date (now December 2027 is being cited in practitioner commentary) but explicitly reaffirms stringent expectations on data and testing governance for high‑risk AI systems, including test data quality, representativeness, and documentation.
Use the extended deadline to turn AI governance into a product‑grade capability rather than a box‑ticking exercise, embedding data and test‑data controls into your MLOps platform and making this a selling point for EU‑exposed customers.
highLeadership may misread the delay as permission to slow down, leaving fragmented logging, undocumented test data, and no auditable AI inventory when large customers or regulators ask for evidence in 2027.
high