OpenLobby.EU
Explore

SaferAI

SaferAI est une organisation à but non lucratif française avec une expertise technique dans la gestion des risques de l'IA.

Lobbying Activity

Meeting with Roberto Viola (Director-General Communications Networks, Content and Technology) and

20 Jan 2026 · Developing European Frontier AI

Meeting with Ann-Sofie Ronnlund (Cabinet of Commissioner Ekaterina Zaharieva)

7 Nov 2025 · AI in Science

Response to Digital package – digital omnibus

14 Oct 2025

This submission is regarding the EU AI Act, though the feedback on harmonised standards could be applicable to other regulations as well. Regarding general-purpose AI (GPAI) models, there is sufficient progress on guidance and compliance mechanisms, in particular the GPAI Code of Practice (CoP), that no delays or changes in the essential requirements on GPAI are necessary. The GPAI CoP is already published after a substantial drafting effort with heavy involvement of GPAI model providers, academia and civil society. And many providers have become signatoriesincluding essentially all of those creating models with frontier capabilitieswhich indicates a strong readiness to comply with the AI Acts requirements. Indeed, the GPAI CoP leverages many existing provider practices, such as Safety and Security Frameworks and Model Reports, simplifying compliance. Regarding high-risk AI systems, the overall scope of the essential requirements for high-risk system providers are reasonable because these AI systems include safety components, remote biometric identification, and medical devices. They pose a significant risk to the health, safety and fundamental rights of those in the EU. The challenge lies in high-risk AI system SME providers achieving compliance with the Act. Harmonised standards are the most direct and efficient way of doing this. As such, the omnibus should provide resources for the drafting of harmonised standards, including support for: participation of more stakeholders, especially those with fundamental rights expertise, and SMEs; officers, and especially secretariats; publically socialising with SMEs the key concepts of the standards being drafted; "welcome packages" for new experts with details on how to write harmonised standards and what the purpose of these standards is. These resources would help ensure harmonised standards granting a presumption of conformity, and the legal certainty this provides, are available sooner. This would especially help SMEs. Additional support can come from the Commission and its ability to clarify the Act by giving guidance on topics of importance to industry. It currently doesnt have the resources to do so however, with important guidance not yet available, such as to answer questions regarding the regulatory boundary of an AI system which is built into a larger technical system under the jurisdiction of other EU laws. Finally, if despite these resources being made available, high-risk AI system SME providers continue to face challenges in compliance with the Acts essential requirements, a time-boxed delay in enforcement (e.g. 6-months) for high-risk AI system SME providers might be warranted, given that they are less able to chart a path to compliance with the essential requirements of the Act in the absence of harmonised standards. Either this, or direct support should be provided from the Commission and national competent authorities to these SMEs to facilitate regulatory compliance questions.
Read full response

Meeting with Ioana-Maria Gligor (Head of Unit Secretariat-General)

22 Sept 2025 · Safe and reliable AI Accelerator

Meeting with Michele Piergiovanni (Cabinet of President Ursula von der Leyen)

22 Sept 2025 · Establishing a joint EU-UK AI control & assurance innovation accelerator - Safer AI in Europe

Meeting with Silvia Bartolini (Cabinet of Executive Vice-President Henna Virkkunen)

17 Sept 2025 · Apply AI and international action on AI for Good

Meeting with Lucilla Sioli (Director Communications Networks, Content and Technology)

2 Sept 2025 · SaferAI proposal to build an Accelerator for projects on AI Safety

Meeting with Ioan-Dragos Tudorache (Cabinet of Executive Vice-President Stéphane Séjourné)

23 Jul 2025 · AI related aspects

Meeting with Silvia Bartolini (Cabinet of Executive Vice-President Henna Virkkunen)

16 Jun 2025 · Apply AI and AI in science

Meeting with Sergey Lagodinsky (Member of the European Parliament)

8 Apr 2025 · Exchange of view