On 27 July 2026, Regulation (EU) 2026/1744 of the European Parliament and of the Council entered into force, more widely known as the Digital Omnibus. This regulation amends the European Union's Artificial Intelligence Act (AI Act), as well as the Machinery Regulation and the Aviation Regulation.
The Digital Omnibus is the first major set of amendments to the AI Act since the regulation was adopted. Its purpose is not to change the content of the AI Act, but to enable a more practical and gradual implementation of its requirements.
What is changing?
High-risk AI systems are given more time. The Digital Omnibus provides that, for high-risk AI systems set out in Annex III of the AI Act, the application of requirements is postponed from 2 August 2026 to 2 December 2027. For high-risk AI systems that are part of regulated products (AI Act Annex I), application of the requirements is postponed until 2 August 2028. This gives organisations additional time to prepare technical documentation, governance processes, and compliance procedures.
A new prohibition on AI-generated illegal content. From 2 December 2026, a prohibition will enter into force on using artificial intelligence to create child sexual abuse material (CSAM), as well as to generate intimate images or videos without the person's consent (deepfake intimate imagery). These provisions supplement the AI Act's existing prohibited-practices rules.
The obligation to label AI-generated content remains. The requirements for labelling AI-generated content remain in force from 2 August 2026. At the same time, the Digital Omnibus provides for a limited transition period for systems already on the market before this date, giving them time to adapt until 2 December 2026.
A more flexible approach to AI literacy. The regulation also clarifies the AI literacy requirement. In place of the previous wording, it is now emphasised that organisations must take the necessary measures to develop AI literacy, allowing more flexibility in choosing the most suitable training and competence-development solutions.
More time for national regulatory sandboxes. Member States have been given a longer period to establish AI regulatory sandboxes — the new deadline is 2 August 2027. This is particularly relevant for Latvia as well, where the Latvian Artificial Intelligence Centre is already running its first AI regulatory sandbox pilot programme.
Stronger powers for the EU AI Office. The Digital Omnibus also clarifies the competence of the EU AI Office, strengthening its role in supervising general-purpose AI (GPAI) models and, in certain cases, where such models are integrated into very large online platforms.
What remains unchanged?
The Digital Omnibus does not change the AI Act's core principles. The risk-based regulatory approach remains in force, as does the regulation of general-purpose AI (GPAI) models, the requirements for systemic-risk GPAI models, and the AI Act's sanctions system. Accordingly, the Digital Omnibus should not be regarded as a "softening" of the regulation, but as a practical revision of the implementation timetable.
PPPA's Commentary
In PPPA's view, the Digital Omnibus gives organisations a valuable opportunity to prepare more thoroughly for the implementation of AI Act requirements.
The additional time should be used not to postpone projects, but to:
- identify the AI systems used within the organisation;
- build an AI governance model;
- define accountability and control mechanisms;
- improve AI literacy;
- integrate AI risk management into existing corporate governance processes.
The AI Act's implementation deadlines have been extended, but the obligation to ensure responsible and safe use of artificial intelligence has not changed.