The EU AI Act was originally presented as the foundation of trust for responsible AI, designed to become a global standard. In this vision, ethics was not an obstacle but a hallmark of European quality. However, with the introduction of the Omnibus VII reform, the narrative seems to have shifted. By choosing to ease certain rules and postpone others, the EU is sending an ambivalent signal: that ethics might be a luxury we can only afford during periods of growth, rather than a sine qua non for technological robustness.

Key Changes Under Omnibus VII

The agreement introduces several critical modifications:

  • Delayed Timeline: The application of rules for high-risk AI is postponed to late 2027 for stand-alone systems and mid-2028 for integrated systems.
  • Relief for Businesses: Regulatory exemptions are extended to Small Mid-Caps (SMCs), and efforts have been made to eliminate overlaps with existing sectoral legislation.
  • Targeted Protections: An immediate ban on non-consensual intimate deepfakes and an accelerated timeline for mandatory AI-content labeling (December 2026).
  • Maintained Transparency: Providers who self-declare as exempt from the “high-risk” category must still register in the EU database, and the AI Office’s supervisory powers have been reinforced.

The Risk of “Compliance Debt”

While this legislative “breather” offers short-term relief for companies, it may have significant long-term repercussions, potentially creating what can be described as compliance debt.

Innovation through Constraint: Industrial history shows that strict standards (whether environmental or safety-related) often act as catalysts for breakthrough innovation. By delaying these requirements, Europe risks incentivizing its companies to develop solutions based on technical standards that may be obsolete by the time the full regulation hits.

The Trust Asymmetry: As citizens and corporate clients demand increasing guarantees regarding data usage and bias mitigation, the absence of a clear legal framework for an additional 16 months could actually stall AI adoption due to fears of reputational risk.

Sovereignty Through Excellence

In the global race for AI development, Europe is not currently setting the pace. However, Europe’s true competitiveness will not lie in its ability to copy deregulated models, but in its capacity to offer “sovereign” technology, technology that is trusted because it is mastered, explainable, and ethical by design.

This simplification of the EU AI Act should be perceived for what it truly is: an opportunity for European companies to use this extra time to prepare for responsible AI usage in a thoughtful and strategic manner.

, , , ,