In recent years the European Union has positioned itself as the world’s foremost digital regulator. From the introduction of the GDPR to the enactment of the Digital Markets Act and the Digital Services Act, the EU has steadily expanded its influence over global digital governance. This phenomenon, often referred to as the “Brussels Effect”, reflects Europe’s ability to shape international regulatory standards beyond its borders.
Recent developments concerning the EU Artificial Intelligence Act (“AI Act”), however, suggest that this regulatory model is now facing significant pressure.
On the 7th of May 2026, Council presidency and European Parliament negotiators reached a provisional agreement on a proposal to streamline certain rules to dilute and delay key portions of the AI Act, particularly those concerning “high-risk” AI systems. The move follows sustained pressure from European industry leaders, who argued that excessive compliance obligations and overlapping regulatory requirements risk undermining Europe’s competitiveness against the United States and China.
While the AI Act remains one of the most ambitious regulatory frameworks for artificial intelligence globally, the recent amendments expose a growing tension within European digital policy: whether the EU can simultaneously maintain rigorous AI oversight while fostering innovation and remaining competitive in the global AI market.
Amendments to the AI Act
The recent provisional agreement substantially alters both the implementation timeline and the scope of several key provisions of the AI Act, particularly those relating to high-risk AI systems.
Under the original framework, many high-risk AI obligations were due to become applicable in August 2026. Following the recent amendments, standalone high-risk AI systems will now become subject to compliance obligations from December 2027, while AI systems integrated into regulated products may not face full compliance obligations until August 2028.
The amendments also narrow the scope of certain obligations. Machinery and industrial systems already governed by equivalent sector-specific safety legislation may now be excluded from parts of the AI Act, reflecting concerns regarding regulatory duplication. This exemption reportedly followed lobbying efforts from major European industrial actors, including Siemens and ASML, who argued that overlapping compliance frameworks would place European manufacturers at a competitive disadvantage.
The rationale underpinning these changes appears primarily economic and strategic. Businesses and several Member States expressed concerns that the cumulative effect of multiple EU digital regulations could discourage investment, slow innovation, and incentivise AI development outside the European Union. The debate therefore shifted from whether AI should be regulated, to how regulation can be calibrated without impairing technological growth.
Importantly, the recent amendments do not represent a total deregulation of AI. Certain politically and socially sensitive provisions have instead been strengthened. The agreement introduces stricter restrictions on AI systems generating non-consensual sexually explicit deepfakes, targets so-called “nudifier” applications, and reinforces watermarking obligations for AI-generated content.
The amendments therefore reveal a more selective regulatory approach. The EU appears increasingly willing to postpone economically burdensome obligations while simultaneously accelerating protections in areas involving fundamental rights, misinformation, and digital harm.
More broadly, the delays may also have significant legal implications for businesses already preparing for compliance. Organisations that invested heavily in AI governance structures based on the original implementation timetable may now face uncertainty regarding future compliance expectations, enforcement priorities, and the interaction between the AI Act and existing sectoral legislation.
The Brussels Effect Under Pressure
The recent amendments may represent one of the clearest challenges yet to the so-called “Brussels Effect”.
Unlike previous digital regulations such as the GDPR, artificial intelligence development depends heavily on computational infrastructure, access to large datasets, rapid innovation cycles, and significant private investment. Excessively burdensome regulation therefore carries the risk of driving both investment and technological development outside the European Union.
The AI Act consequently exposes a growing constitutional and economic tension within EU law, with the attempt of balancing the protection of fundamental rights, including privacy, non-discrimination, and human dignity, against the need to promote innovation, competitiveness, and technological sovereignty.
The recent dilution and postponement of certain provisions suggest that the EU may be shifting away from a strictly precautionary approach towards a more economically pragmatic model of AI regulation.
Conclusion
The recent dilution and postponement of portions of the EU AI Act mark a significant moment in the evolution of European digital governance. The EU now appears to recognise that regulating artificial intelligence is not solely a legal exercise, but also a strategic and economic one.
For Malta, these developments will have important implications for businesses operating within regulated sectors, technology providers, financial services operators, and organisations integrating AI systems into their operations. As the regulatory landscape surrounding artificial intelligence continues to evolve across the European Union, our firm will continue monitoring developments closely and advising clients on the legal and regulatory implications of the AI Act.
The information provided in this Insight does not, and is not intended to, constitute legal advice. All information, content, and materials available are for general informational purposes only. This Insight may not constitute the most up-to-date legal information and you are advised to seek updated advice.
