August 2, 2026 is the date when the EU AI Act’s Annex III high-risk provisions become applicable to providers and deployers of covered AI systems. Per the European Commission’s official regulatory framework page, this milestone falls 24 months after the Act’s entry into force on August 1, 2024. Sixteen weeks remain.
The Act entered into force on August 1, 2024. General provisions, including AI literacy obligations, became applicable February 2, 2025, per the EU AI Act Service Desk’s implementation timeline. The August 2, 2026 date is the next major threshold, covering high-risk AI systems defined under Annex III of the Act. Full implementation of remaining provisions follows on August 2, 2027.
Who the August 2 deadline affects.
Annex III covers eight categories of high-risk AI systems: biometric identification and categorization, critical infrastructure management, education and vocational training, employment and worker management, access to essential private and public services, law enforcement, migration and border control, and administration of justice. If your organization provides or deploys AI systems in any of these categories to EU markets or EU residents, the August 2, 2026 deadline applies.
Providers of general-purpose AI models also have obligations under Title VIII, those timelines overlap with the high-risk provisions for certain model classes.
What “applicable” means in practice.
Applicability isn’t notification. It’s enforcement eligibility. On August 3, 2026, national market surveillance authorities can begin assessing penalties for non-compliant high-risk systems. Penalties for high-risk system non-compliance can reach up to €15 million or 3% of total worldwide annual turnover, per EU AI Act Article 99(4). Note: This figure is drawn from training knowledge of EU AI Act Article 99; verify against the official text at eur-lex.europa.eu before citing in legal or compliance contexts. Organizations with prohibited AI systems already faced the harder deadline, February 2, 2025, when prohibitions took effect, carrying penalties up to €35 million or 7% of global turnover.
One provision compliance teams often miss.
According to legal analysis from Orrick, AI systems already placed on the market before August 2, 2026 may benefit from an extended compliance period. If your organization deployed a covered AI system before the deadline, the specific grandfathering terms applicable to your situation require verification with legal counsel, the provision isn’t a blanket exemption.
What to watch.
The EU AI Office is publishing guidance and technical standards through 2026. The compliance deadline calendar on this page tracks upcoming regulatory milestones. Organizations waiting for final harmonized standards before beginning conformity assessment should note that waiting eliminates buffer time for remediation if gaps are found.
TJS synthesis.
Sixteen weeks is enough time to start. It isn’t enough time to start from scratch and finish. Organizations that haven’t completed their Annex III system inventory and begun conformity assessment are now in catch-up mode. The first step is knowing which systems you operate that fall within Annex III categories, that determination alone takes time and requires legal input.