On 2 August 2026, Europe's grace period on its landmark AI law ends a…

On 2 August 2026, the EU AI Act's grace period expires and the European Commission gains enforcement powers to fine makers of the most powerful AI models for non-compliance with training disclosure requirements and systemic risk assessments. The law applies a lighter regime to all general-purpose AI providers and a heavier tier of self-policing duties to roughly a dozen 'systemic risk' models trained with over 10^25 floating-point operations, with penalties up to 3% of global turnover. If you're building one of the dozen models that cross the 10^25 compute threshold, you're about to get audited by a regulator with power to pull your system off the European market or fine you 3% of global revenue. The disclosure rules have gaps—the Commission doesn't audit training data itself, only acts on complaints—which means you own the credibility risk if your disclosures don't hold up.