Europe readjusts AI rules to make them easier to apply
Articles9 June 2026
Key aspects of the agreement reached in the European Union to revise the IA Regulation, focused on improving its practical applicability by adjusting deadlines, simplifying obligations and specifying certain technical requirements.
The European Union has reached a provisional agreement to partially amend the Artificial Intelligence Regulation ("ARI") with the aim of simplifying its application, reducing regulatory burdens and better adapting it to technological reality. While maintaining the basics of the RIA, the reform introduces new deadlines, technical adjustments and a specific ban on particularly harmful uses of AI.
Key points:
- Delay in obligations for high-risk AI: rules applicable to high-risk AI systems (such as biometrics, employment, education or border control) are postponed until 2 December 2027, giving more time to companies and regulators.
- Extended deadlines for AI embedded in products: AI systems embedded in products subject to sectoral regulation (such as devices or machinery) will have to comply from 2 August 2028.
- Prohibition of nudifier applications: an express prohibition of AI systems generating non-consensual sexual or intimate images or child abuse content is introduced, applicable from 2 December 2026, reinforcing the protection of users.
- Postponement of transparency obligations (watermarking): rules for identifying AI-generated content (such as images or videos) are delayed until 2 December 2026, to allow for the development of appropriate technical standards.
- Regulatory simplification and business support: the agreement aims to reduce administrative burdens, provide more legal clarity and facilitate compliance, especially for SMEs and growing businesses.
In this regard, the agreement reached must complete its processing in accordance with the European Union's legislative procedure for its final adoption, which is expected before 2 August 2026.
Article written by the TMT area of ECIJA Madrid.
Links:
