EU lawmakers are reportedly weighing a delay in the application of key provisions under the bloc’s Artificial Intelligence Act (AI Act), raising questions over the enforcement timeline of one of the EU’s most ambitious digital regulations.
Although the AI Act officially entered into force on 1 August 2024, its obligations are staggered, with different provisions taking effect in phases over a three-year period. The first set of requirements—prohibiting certain AI practices and mandating AI literacy initiatives—have been in force since 2 February 2025. However, doubts are emerging over whether subsequent obligations will be enforced as scheduled.
These include obligations related to general-purpose AI (GPAI) models, due to apply from 2 August 2025, as well as transparency rules and specific requirements for high-risk AI systems listed in Annex III, which are scheduled for 2 August 2026. A further set of provisions for high-risk AI systems under Annex I is expected to take effect on 2 August 2027.
According to several sources familiar with ongoing discussions, the European Commission is now actively considering postponing the next phase of implementation. While no official announcement has been made, the idea of delaying the enforcement timeline has gained traction, particularly under the current Polish Presidency of the Council of the EU. Poland has reportedly submitted proposals advocating a delay to the Act’s upcoming application dates.
For 🇵🇱 Deputy Minister @DStanderski, the enforcement of the AI Act should be contingent upon the finalization of compliance tools, including the code of practice for GPAI models. However, the potential postponement should also be subject to conditions.https://t.co/rjhD1IU6tx
— Luca Bertuzzi (@BertuzLuca) June 11, 2025
Delays Extend Beyond the AI Act Text
The delay in enforcement is not limited to the statutory obligations of the AI Act. Article 56(9) of the regulation had originally required the publication of a code of practice for GPAI models by 2 May 2025. This deadline has already passed without the code being finalised, due in part to stakeholder concerns over the direction of early drafts. The document was expected to provide practical guidance on implementation, especially for large-scale AI developers and deployers.
In addition, the technical standards that are meant to support compliance have not yet been published. These standards are necessary to enable companies and public bodies to assess their systems and demonstrate conformity with the regulation’s requirements. The absence of these reference points adds further uncertainty to the feasibility of the 2 August 2025 deadline.
Precedents for Delay and Legislative Options
While the Commission has not outlined how any deferral would be enacted, several precedents suggest that a formal delay could be introduced rapidly through secondary legislation. In 2024, for example, the EU adopted a regulation to postpone the implementation of the Deforestation Regulation by one year. More recently, in 2025, a directive delaying the application of the Corporate Sustainability Reporting Directive and the Corporate Sustainability Due Diligence Directive was proposed, adopted, and published within seven weeks.
Legal observers suggest that a similar mechanism could be used to delay the enforcement of GPAI obligations under the AI Act. Options include the adoption of a standalone directive or an amending regulation. Either would need approval from both the Council of the EU and the European Parliament.
Timing is a critical factor. The GPAI obligations are scheduled to take effect on 2 August 2025. For any postponement to be implemented in time, an agreement would have to be reached before the European Parliament’s summer recess. The final plenary session before the break is during the week commencing 7 July. This means that both institutions would need to conclude negotiations by 10 July if a delay is to be formalised ahead of the August deadline.
Political Will Remains a Deciding Factor
While the procedural route for a delay exists, its success hinges on whether there is sufficient political consensus. Some Member States and stakeholders have raised concerns about the readiness of the ecosystem to meet the AI Act’s obligations on time. Others argue that the urgency of establishing clear AI rules—particularly for GPAI models—requires swift implementation, not postponement.
There are also implications for international regulatory alignment. The EU has positioned the AI Act as a potential global benchmark, with regulators in jurisdictions including Canada, Brazil, and the United States observing its progress closely. A delay could affect the perceived credibility of the bloc’s regulatory leadership in the AI domain.
Nonetheless, with the absence of key compliance tools such as technical standards and a finalised code of practice, there appears to be growing recognition within the institutions that immediate enforcement of the next phase of obligations may not be practical.
Read also:
EU AI Act: member states reach landmark deal on AI Regulation
