As the EU's AI Act takes effect in August, critics say the law is undercooked and regulators left out essential details needed to give clarity to businesses
Javier Espinoza / Financial Times :
Context & Ripple Effects
The Act moved from a disputed final-negotiation phase, including pressure for a foundation-model carve-out, to publication of its final text ahead of its August 1 entry into force. That sequence matters because the rulebook’s arrival starts the compliance clock even as critics argue that practical guidance remains incomplete.
Related coverage says most companies have until August 2026 for key transparency compliance, creating a gap between formal enactment and the operational detail businesses need to implement the regime.
First-order effects
- AI developers and companies deploying covered systems must begin assessing which obligations and deadlines apply, despite criticism that regulators have not supplied enough implementation detail.
- Unclear guidance raises immediate legal and planning risk for businesses trying to design compliance programs before the main requirements become applicable.
Second-order effects
- Legal, audit and AI-governance providers gain a more central role as companies seek interpretations of a framework whose final text is published but whose application remains contested.
- Ambiguity can delay product and deployment decisions at the margins, as firms weigh the cost of building to a stricter interpretation against the risk of later enforcement.
Third-order effects
- The Act’s credibility will depend not only on statutory rules but on whether regulators can translate them into sufficiently predictable implementation guidance; persistent uncertainty would reinforce later criticism that the framework privileges regulation over innovation.
- The episode points to a broader shift toward compliance becoming a product-design constraint for AI systems, especially where rules differentiate among model and deployment risks.
The trend: AI governance is shifting from legislative bargaining to the harder task of turning broad risk-based rules into operational standards that companies can reliably follow.