It agreed early Thursday to a major overhaul of the regulatory framework for , postponing for more than a year the application of restrictions on high-risk applications. The decision came after intense pressure from governments and industry, in a negotiation that lasted until the early hours of the morning.
The agreement concerns the EU’s flagship artificial intelligence law, which the European Commission, the European Parliament and member states had drawn up to regulate the most dangerous uses of the technology. The new arrangements postpone the full implementation of the rules to December 2027, while also introducing exemptions and transitional periods for critical sectors.
Pressure from industry and governments is changing the pace of implementation
Central to the agreement is the exclusion of much of the industrial application of artificial intelligence from the scope of the law, a development seen as a victory for Germany and its industrial powerhouses such as Bosch.
The new regulation allows industrial AI uses to mainly fall under an already existing framework for machines and industrial equipment, avoiding a double regulatory burden. The decision pleased Berlin, which had warned that tougher rules would put European companies at a disadvantage against competitors from the US and Asia.
In contrast, areas such as medical devices remain within the scope of the AI Act, which maintains high levels of regulatory scrutiny over critical applications.
The transitional period and new obligations for companies
The deal sees the rules on high-risk uses take full effect in December 2027, giving businesses extra time to adjust.
At the same time, the adaptation time for the obligations of marking content produced by artificial intelligence, such as watermarks on images and videos, is reduced. The grace period is limited to three months, instead of the six originally proposed.
Tougher stance on deepfakes and child pornography
Despite relaxations in the regulatory framework, the agreement maintains and strengthens critical prohibitions. Particular emphasis is placed on artificial intelligence systems that produce sexualized deepfakes of recognizable faces, which are outlawed.
Accordingly, the use of AI to create child pornography material is prohibited, in an effort to address increasing incidents of malicious use of generative technologies.
Political balance between innovation and regulation
The deal marks the EU’s first substantial rollback in digital law, at a time of heightened geopolitical and economic pressures. The European Union is trying to strike a balance between the need to protect citizens and maintain the competitiveness of the European technology industry.
European Commission President Ursula von der Leyen stressed that the agreement creates a “simple and innovation-friendly environment” for artificial intelligence in Europe, while underlining the strengthening of citizen protection.
At the political level, this development also reflects the EU’s attempt to avoid excessive regulation in an area where the US and China are moving faster, with the stakes of technological primacy remaining open.
