Brussels pivots: EU scales back landmark AI regulation

The amendments arrive as the European Commission grapples with internal delays and mounting pressure from industry lobbyists. By postponing rules for high-risk AI applications until December 2027—and pushing sector-specific requirements to 2028—Brussels aims to grant companies a longer compliance runway. A major win for German Chancellor Friedrich Merz, the new industrial exemption ensures that car manufacturers and chip-lithography firms remain largely shielded from the law’s most restrictive oversight.

Despite these concessions, the framework for general-purpose models, such as those from OpenAI and Anthropic, remains on track for an August enforcement date. Proponents of the changes, including tech policy lead Vicente Domecq, argue these adjustments are essential to prevent Europe from falling further behind in the global AI race. Yet, critics contend the shift is a response to intense lobbying by U.S. tech giants, noting that the EU AI Office has struggled to define technical standards or staff its teams with the expertise necessary to manage the law’s original, more ambitious scope.

With the upcoming August deadline for general-purpose AI, the focus now turns to whether the EU can balance its regulatory ambitions with the rapid evolution of the technology. As shadow rapporteur Axel Voss noted, the current framework may already be aging, forcing policymakers to reconsider whether they can build the institutional capacity to regulate at speed without collapsing under the weight of their own bureaucracy.