European Union legislators agreed Thursday to postpone critical restrictions on high-risk artificial intelligence uses by more than a year, while granting sweeping exemptions to industrial AI applications in a deal that marks the first significant rollback of digital protections in the bloc. The agreement, reached after marathon negotiations that stretched until 4:30 a.m., came after heavy pressure from industry and national governments.
The Cypriot presidency of the Council of the EU and the European Parliament confirmed the deal, which delays restrictions on high-risk AI until December 2027. Originally, rules governing high-risk uses were set to kick in this August under the EU's AI Act, which became law in August 2024 after years of negotiations designed to protect citizens from algorithmic harms.
Industrial AI Gets Special Treatment
The agreement largely exempts the use of AI in industrial applications from the scope of the law, sparing companies from having to comply with AI requirements under the legislation. Instead, these firms will only need to meet AI standards under separate machinery rules. The change represents a major victory for Germany, where Chancellor Friedrich Merz and other top officials lobbied intensely to keep tech giants Siemens and Bosch competitive by avoiding what they called a double regulatory burden.
EU countries had backed Germany's demand, though negotiators confirmed that other industries under discussion, including medical devices, were not exempted and will still be covered by the AI law. The selective carve-out raises questions about which workers and consumers will benefit from AI protections and which will be left vulnerable to unchecked algorithmic decision-making in their workplaces and daily lives.
Pressure From Industry and Washington
The deal reflects mounting pressure on the EU from the United States over its tech laws, as well as warnings from the bloc's own industry and governments that strict restrictions had put Europe at a disadvantage in a global AI race. Commission President Ursula von der Leyen defended the agreement, saying it "provides a simple, innovation-friendly environment" for AI in Europe while strengthening protections for citizens. "For safe and simple AI governance in Europe," she said on X.
Critics, however, see the rollback as evidence that corporate competitiveness concerns are trumping the safeguards that lawmakers originally promised would protect workers, consumers, and vulnerable communities from algorithmic discrimination and other AI-related harms.
Some New Protections Added
The deal does include some enhanced protections. It bans AI systems that can generate sexualized deepfakes of intimate parts of "identifiable" people, responding to global outrage over the abusive use of Elon Musk's AI tool Grok. AI systems that generate child pornography will also be banned under the agreement.
Companies will receive a grace period on meeting new requirements to watermark AI-generated content, though that grace period will only be three months rather than the six months originally proposed. The shortened timeline suggests some resistance to giving industry unlimited flexibility to delay transparency measures.
Why This Matters:
The rollback of EU AI protections reveals a fundamental tension between innovation and accountability in the regulation of powerful new technologies. By delaying high-risk AI restrictions until December 2027 and exempting industrial applications, the agreement leaves workers in manufacturing and other sectors exposed to algorithmic management systems without the safeguards lawmakers promised when the AI Act became law in August 2024. The selective exemptions raise concerns about unequal protection, where some industries face oversight while others operate with minimal constraints. As the EU faces pressure from both Washington and its own corporate champions, the agreement suggests that the bloc's commitment to human-centered technology governance may be weakening at precisely the moment when AI systems are becoming more pervasive in employment, healthcare, and public services. The decision sets a precedent that could embolden further industry demands to water down digital protections.