Legal development

Digital Economy Soundbite | What’s up with AI and the EU Digital Framework?

Close-up of an element

    The digital rulebook in Europe is a tangled web, particularly where AI meets other digital legislation.  A recent study published by the European Parliament finds that the frequent overlaps between the AI Act and adjacent regulatory instruments deter AI uptake, delay time to market and introduce compliance asymmetries across Member States. In its current state, the complex network of differing obligations places a disproportionate burden on EU AI innovators, especially domestic SMEs and start-ups who remain ill-equipped to absorb compliance costs.

    Interplay with EU Digital Legislative Framework

    The study highlights that the overlay of the AI Act imposes varying obligations on the same actors, sometimes producing duplicative, inconsistent, or unclear requirements. AI deployers must therefore carefully assess how the AI Act’s scoping, risk categorisation and conformity regime sits alongside horizontal and sectoral regulatory frameworks. Navigating this legal framework is not simple and presents three major pressure points:

    • Overlapping regulatory duties and assessments burdens already cumbersome and resource intensive compliance frameworks.
    • The introduction of the AI Office introduces yet another enforcement entity into an already fragmented mix of regulators. Differing data protection and market surveillance bodies may find themselves running overlapping investigations or issuing duplicative information requests.
    • Finally, inconsistent obligations contained in separate regulations, combined with a lack of accompanying guidance or precedents, leave AI providers and deployers with no clear signposts as to how they can reconcile conflicting duties.  

    Recommendations

    The study concludes that much work will need to be done to ensure simplification and thus promote innovation. Key recommendations include:

    • Short term: Advance coordinated guidance and aligned enforcement among supervisory authorities. Support cross-recognition of compliance processes and establish harmonised sandbox frameworks.
    • Medium term: Pursue targeted and minimal legislative amendments to clarify key definitions and reduce duplicative obligations.
    • Long term: Reassess the EU’s digital regulatory framework to promote consolidation, simplification, and strategic consistency. Move away from fragmented, technology specific  instruments and streamline stakeholder role classifications. Enable an integrated regime that delivers agile compliance for innovators while safeguarding the Union’s commitments to fundamental rights and safety.

    What now for the EU digital rulebook?

    Brussels has already acknowledged the need to simplify the EU digital legislative framework with the upcoming Digital Omnibus. This is taking place against a backdrop of wider political calls for deregulation, such as the European Parliament's Committee on Economic and Monetary Affairs (ECON) draft report and motion cautioning against an overly restrictive regulatory approach to AI deployment, fierce pressure from the Trump administration to ensure that EU digital legislation does not overly restrict US tech, and industry calls to stop the clock on the AI Act itself. More recently, press speculation on proposed AI Act amendments contained in the Digital Omnibus has heightened. Though a final version is expected to be published later this month, current thinking suggests targeted amendments to enable smoother, proportionate implementation of the AI Act, including:

    • an additional one-year  grace period for "watermarking" obligations;
    • extending SME privileges, such as simplified documentary requirements or special rules about penalties, to small mid caps;
    • more flexibility on AI literacy with the Commission and member states playing a bigger part;
    • easing post market monitoring and registration for AI systems used in high-risk areas;
    • permitting providers and deployers of AI to process special category  data for bias correction;
    • expanding sandboxes and real-world  testing;
    • centralised oversight of AI embedded in very large platforms and search engines with the AI Office; and
    • clarification on the interplay between the AI Act and other EU legislation.

    Meanwhile, businesses and regulators continue to struggle with uncertainty. 

    The information provided is not intended to be a comprehensive review of all developments in the law and practice, or to cover all aspects of those referred to.
    Readers should take legal advice before applying it to specific issues or transactions.