GSMA和Connect Europe对数字综合人工智能监管提案的看法

As AI becomes the backbone of next-generation connectivity, its successful integration into electroniccommunicationsnetworks(ECN)will determine Europe’s ability to lead in innovation,competitiveness, and digital sovereignty. Europe stands at a critical juncture in shaping the future of AI regulation. AsAIbecomes integral tonext-generation connectivity, the EU must ensure that its regulatory framework supports innovationwhile safeguarding competitiveness and digital sovereignty. In this context, Connect Europe and the GSMA, representing the European telecommunicationssector, welcome the European Commission’s proposal for a Digital Omnibus on AI rules as a positivefirst step towards ensuring the effective implementation of the AIAct, while supporting theCommission’s goals of competitiveness and simplification.We believe several proposals rightlycontribute to the simplification objective including the changes proposed to the literacy obligations orthe (non) registration of non-high risk AI use cases. However, for this proposal to succeed,several considerations must be addressed. We call for furthersimplification of the implementation deadlines for high-risk AI provisions and a swift adoptionof theproposalto grant legal certainty to stakeholders. Connect Europe and the GSMA call on policymakers to deliver clarity, proportionality, and realistictimelines to enable compliance and foster growth. Without decisive action, fragmented deadlines andlegal uncertainty risk slowing progress and underminingEurope’s ambition to lead in AI-driventransformation. 1.High-risk rules should be implemented one year after standards are made availableConnect Europe and the GSMAwelcome the proposal to postpone the implementation timeline of high-risk rules due to the delays in the availability of standards.These standardsand guidelinesarecriticalto provide legal certainty, support compliance efforts, and uphold proportionality. However, implementation and internal compliance processes can only start once these standardsandguidelinesarefinalized,andsufficient time should be allowed for these processes to ensure they arecarried out thoroughly and effectively. For this reason,standards related to the high-risk provisions of the AI Act must be in place at leastoneyearbefore the provisions are implemented. The current proposal–which grantsa6-monthtransitionperiod forAnnex III systems–will not provide companies with sufficient time to prepare for theimplementation. Therefore,for simplification and legal certainty,the proposal should delay implementation for oneyear after standardsand guidelinesare in place, in line with the Commission’s previously expressedconsiderations, andthe calls ofmany stakeholders. 2.Implementation datealignmentfor Annex I and Annex IIIThe AI Act currentlyforeseesdifferent deadlines forthe transition period for high-risk rules related to Annex I and Annex III of the AI Act. Inlight of further simplification efforts and proposed changes tothe original timeline, wecall foraligning these timelinesto ensure clarity and consistency for allstakeholders.Thismeans that both Annexes I and III cut-off date will be August 2028.Suchharmonisationwould reduce administrative burden, provide certainty for industry planning, andsupport a smoother, more efficient rollout of the AI Act. 3.TheDigital Omnibus must be approved in timeIt isessentialthat the Digital Omnibuson AI rulesis approved as quickly as possible.Withthe original implementation deadline for high-risk provisions related to Annex IIIset for 2 August 2026, there is amajor risk thatthe Digital Omnibus is not approvedin time.We therefore call for all policy makers toprioritizethe AI Digital Omnibus to ensure its timely approval and publication toavoidcreating legaluncertaintyfor all market playersthatcould derail the overall implementation of the AI Act.As aconcrete solution to avoid legal uncertainty regarding the timelines for Annexes I and III we suggestthat the provisions regarding application of requirements for high-risk AI systems should be proposedseparately from the other parts of the AI Omnibus, which may require additional deliberation andnegotiations between the co-legislators. ConclusionsThe European Commission’sproposed actionscan play a decisive role inensuringthe smooth implementation ofAI Act’s high-risk provisions, provided they are, formulatedwith clarity andbalance. Torealizethis potential,swift approval withalignment between Annexes I and III leading toclearerandmoreproportionate implementation datesisessential. Thisis notmerelya regulatorynecessity, it isastrategicimperative for the EU. By fosteringAI-powered innovation in Europe’s vitalindustries, such as telecoms,rather thanallowinglegal uncertaintyto hinder progress, Europe canstrengthen its global competitiveness and secure its leadership in the digital era.