August 2 is 26 days away. Only 8 of 27 EU states have AI enforcement authorities. The regulation your compliance team is racing toward has no one to enforce it.

August 2, 2026.

That's when the EU AI Act's high-risk provisions go live.

Penalties: up to €35 million or 7% of global annual turnover. Exceeds GDPR.

Here's the problem your compliance team isn't tracking:

Only 8 of 27 member states have designated national enforcement authorities. France, Germany, and the Netherlands haven't finished the legislative process. The bodies that are supposed to police your AI systems don't exist yet.

The technical standards from CEN/CENELEC — the benchmarks you need for conformity assessments — won't be published until Q4 2026 at the earliest. The first standard only entered public enquiry in October 2025.

The Digital Omnibus proposal would push high-risk compliance to December 2027. Parliament approved it June 16. Council approved it June 29. But it hasn't been published in the Official Journal. Until it is, the original August 2 deadline is legally binding.

78% of organizations haven't taken meaningful compliance steps. First-year costs for large enterprises: €8-15 million per system.

If your AI systems touch EU users — hiring tools, credit scoring, biometric ID, customer-facing automation — you're exposed. Extraterritorially.

Audit your AI inventory today. Classify every system by risk tier. Document good-faith compliance efforts now. The enforcement infrastructure is broken. Your compliance strategy can't afford to be.