26 submissions from 26 organizations told the European Commission what they think about this file. Here is what each of them said, in their own words.
The Commission lists 51 submissions on this file. Shown here: the 26 from organizations. Not shown, by design: submissions from private individuals, which we never publish, and anything filed since our last weekly refresh.
Who showed up
14 submissions from industry — companies and their trade associations — against 7 from civil society: NGOs, consumer organizations, environmental groups and trade unions. That is 2 industry submissions for every one from civil society.
Industry 14Civil society 7Public authorities, academia, other 5
Groupings use the respondent type each organization selected when filing. Counting submissions, not organizations — a body that filed twice is counted twice.
What the room declares
17 of 26
in the EU Register
1,063
full-time lobbying staff
€7.8M+
declared costs a year
27
EP accreditations declared
Self-declared to the EU Transparency Register (snapshot 30 Aug 2026). The cost figure sums band floors, so the true total is higher.
The file, right now
The consultation closed on 9 Apr 2026 — it ran from 12 Mar 2026.
Policy area
Digital & tech (DG CNECT)
Where it stands
Awaiting adoption
Adoption expected
30 Jun 2026
How it got here
Draft implementing regulation9 Apr 2026
Also on the Commission’s pipeline for this file, with no date recorded: Initiative planned, Implementing regulation.
The proposed implementing regulation states that what is called AI is in fact an evolved search engine that must not be influenced by the topics it deals with, I found it: (1) it is difficult to establish a proven relationship between the motive behind the murder of Edoardo Agnelli and his nephew J Elkann who took control of the Agnelli Group because Gianni Agnelli’s donation to his son was cancelled by his murder.
Filed in Italian · English published by the European Commission
Pour Demain welcomes the Commission's draft Implementing Regulation as a meaningful step toward enforcing GPAI model governance under the AI Act, in particular regarding specific language on appropriate model access. Building on this strong foundation, our recommendations focus on three areas of improvement from a safety and security perspective.
The Future of Life Institute (FLI) is an independent nonprofit organisation with the goal of reducing large-scale risks and steering transformative technologies to benefit humanity, with a particular focus on artificial intelligence (AI).
The Computer & Communications Industry Association (CCIA Europe) welcomes the possibility to provide feedback on the European Commissions draft implementing regulation on detailed arrangements for the conduct of certain proceedings by the Commission pursuant to the AI Act. We respectfully submit our observations in the attached document.
Please find attached ITI's contribution to the consultation on the draft implementing regulation on evaluations and proceedings for General Purpose AI (GPAI) Models under the AI Act. ITI - the Information Technology Industry Council, is the global trade association of the tech industry, representing 80 of the worlds most innovative companies operating in the software, hardware, services and related industries.
Across all areas, we have high level concerns on: timelines, access to source code, and confidentiality. This submission covers four articles only in substance: Arts 2, 7, 9, and 14. The scope of access granted to the Commission in Article 2 is overly broad, and the timelines and mechanisms through which this access will be granted in Article 7 and 14 are not realistic from a business perspective.
AI Standards Lab welcomes this draft Implementing Regulation for Article 101 of the AI Act as a well structured procedural instrument for the evaluation of general-purpose AI models and enforcement proceedings under the AI Act. Find enclosed a 1 page document with a few brief considerations and recommendations, including on: 1.
GENERAL The Czech Republic has concerns regarding provisions that may impose disproportionate burdens on providers of general-purpose AI (GPAI) models, endanger their intellectual property or exceed the scope of the empowerment granted by the AI Act. The concerns set out below are listed in order of priority.
Ladies and gentlemen, We would like to thank you for the opportunity to submit comments on the Implementing Regulation GPAI Enforcement. Given the short deadline for an opinion, our members could not be consulted separately.
Filed in German · English published by the European Commission
In its current form, the proposal is strong as an investigative and enforcement tool, but weaker as a framework for procedural equality and democratic scrutiny. In terms of fairness and competitiveness, the greatest risk is that the rules are practically best managed by the largest global players, while smaller European players bear relatively higher costs.
Filed in Czech · English published by the European Commission
BSA welcomes the opportunity to comment on the European Commissions public consultation on the Draft Implementing Regulation on detailed arrangements on evaluations and proceedings with regards the Artificial Intelligence Act.
EUROCINEMA welcomes the Commissions efforts to establish detailed procedural rules for the evaluation of general-purpose AI models and the conduct of proceedings under Regulation (EU) 2024/1689. In particular, the clarification of access powers under Article 2 and the framework for the appointment of independent experts under Articles 3 and 4 constitute important steps toward ensuring effective supervision.
EURAMET, the European Association of National Metrology Institutes, welcomes the opportunity to contribute to the European Commissions (EC) open consultation on Artificial Intelligence Act detailed arrangements on evaluations and proceedings. Measurement science (metrology) is crucial in underpinning accuracy and standardisation.
With regard to the proposed Implementing Regulation, we appreciate that the Commission has taken this step to clarify how Regulation (EU) 2024/1689 will be applied. However, from the manufacturers perspective, we believe it would be very helpful to have more detail on certain key aspects, particularly regarding implementation timelines, including clear milestones that would enable us to organise and plan our…
The Actuarial Association of Europe (AAE) welcomes this Implementing Regulation as a well-structured procedural framework for the evaluation of general-purpose AI (GPAI) models and enforcement proceedings against GPAI providers. We recognise that, as an implementing act, the scope for substantive change is limited, and our comments are accordingly focused and targeted.
Given the speed of modification of the systems generated by Dallia, it is obviously necessary to update the European AI Act 2024. It will be necessary to continue on the path of a risk-based approach with a strong focus not only on technological regulation, but above all on making use of it.
Filed in Italian · English published by the European Commission
OpenMined welcomes the draft implementing regulation as a useful step in clarifying how the Commission will conduct evaluations of general-purpose AI models under Article 92 of the AI Act. The draft would, however, benefit from stating more explicitly that, where appropriate, evaluation access should follow a proportionate and least-intrusive approach before resorting to direct access to source code, model weights…
Re: Draft implementing regulation evaluations (Article 92) and proceedings (Article 101) under Regulation (EU) 2024/1689 The draft specifies how the Commission will evaluate a general-purpose AI model the specific trained parameters that generate the system's outputs, which can be changed independently of the service that hosts them and how it will conduct proceedings that may result in fines.
The German Association of the Automotive Industry (VDA) welcomes the objective of the European Commission to establish, through the present Implementing Regulation, a uniform procedural framework for measures pursuant to Regulation (EU) 2024/1689 (AI Act).
Filed in German · English published by the European Commission
The effectiveness of the Artificial Intelligence Act will depend not only on its substantive obligations but also on the clarity, fairness and robustness of its enforcement framework. The proposed detailed arrangements on evaluations and proceedings are therefore a critical component in ensuring that the regulation can be applied consistently and credibly across the Union.
Connect International welcomes the draft Implementing Regulation and the European Commissions efforts to establish clear and enforceable procedural arrangements for the evaluation of general-purpose AI (GPAI) models. However, Article 3 creates a striking imbalance: it is highly specific in defining what an expert must not be, while remaining largely silent on what an expert must know.
We welcome the draft Implementing Regulation and its efforts to provide procedural clarity for the evaluation of general-purpose AI models and related proceedings. We would like to highlight several points for consideration: Interdisciplinary expertise (Articles 34) We recommend that the selection of independent experts explicitly ensures interdisciplinary representation, including experts from civil society, media…
1. Clarity of Access Requirements (Article 2) The document lists many possible forms of access to a generalpurpose AI modelAPIs, internal interfaces, source code, model weights, infrastructure, and system-state interactions. While the scope is clear, the text does not explain when each type of access is appropriate or required.
The draft provides a clear procedural framework for access, evaluation, and proceedings. However, its operationalisation assumes that access and evaluation can be performed without a standardised technical substrate.
Feedback on the Draft Commission Implementing Regulation (Ares(2026)2709234) Article 2 Model access scope is overbroad. The phrase "all levels of access granted to employees" sets an inconsistent, provider-dependent standard.
Future Generations & Future Narrative Labs plus Oral Health Foundation & Iaomt plus BSc Dental Technicians World Wide Will Via municipalities & diarienumbers & tickets empower best treatments with no side effects as insurance of NegarTime via @etermalnafas & @wipo_nafas when we apply for WIPO Global Award Entry 2026 as TechConnect 730 Abstact when I asked professor Alberto Alemanno to review my innovation via AI…
Method. Every quote is verbatim from the organization’s own submission to the European Commission, trimmed to its opening passage and never summarized by a model. Where a submission was filed in another EU language we show the English text the European Commission publishes alongside it, labeled on the quote; the original is one click away at the source. Groupings use the respondent type the organization itself selected when filing. We deliberately do not label anyone “supportive” or “opposed” — you read what they wrote and draw your own conclusion. Organizations only, never individuals. Reused under Commission Decision 2011/833/EU; the European Commission is not liable for this reuse.
