Andrijko Z. / Wikimedia Commons (CC BY-SA 4.0)
The EU AI Act's grace period is over — Brussels can now hand out fines
As of August 2, 2026, the EU AI Act's one-year grace period has ended. The European Commission can now fine the makers of the most powerful AI models up to 7% of their global annual turnover.
Nova AI News Editor
August 10, 2026 · 2 min read
August 2, 2026 marked a symbolic threshold for the EU AI Act: the one-year grace period granted for the law's provisions covering the most powerful general-purpose AI (GPAI) models officially ended. From this date, the European Commission has direct power to fine companies that don't comply.
How big are the fines?
The law sets out a tiered penalty structure:
- Up to €35 million or 7% of global annual turnover for prohibited practices
- Up to €15 million or 3% of turnover for breaches of other duties under the Act, including transparency obligations
- Up to €7.5 million or 1% of turnover for providing incorrect, incomplete, or misleading information
For a large tech company, a 7% fine could run into billions of euros — making this the most heavily sanctioned AI regulation anywhere in the world so far.
Who's in scope?
The Commission's focus is on companies building models deemed to carry "systemic risk." Compliance guides estimate around a dozen models globally fall into this category. Companies like OpenAI and Anthropic fall under the new scrutiny even if they're based outside the EU — as long as they place systems on the EU market.
That's a significant shift for US-based labs: making a model available in the EU now means being accountable to Brussels for its training data, safety testing, and real-world behavior.
What are companies doing?
Compliance guidance recommends that in-scope companies prepare documentation covering: transparency of data sources used in training, copyright compliance, systemic risk assessments, and incident reporting processes. Several labs announced dedicated EU compliance teams in the final months of the grace period.
Why it matters
Until this date, the EU AI Act was largely a framework on paper. Now backed by real financial enforcement power, it will directly shape both how major labs approach offering models in Europe and the broader global debate over whether AI development converges on the EU standard or fragments into region-specific versions.
Related Articles
Comments
No comments yet — be the first to comment.