14/10/2026
On Wednesday 14th of October, Dublin hosts an invitation-only AI summit where executives of companies and business groups hold 82% of the speaking slots, officials and a single trade unionist share the rest, and civil society holds none.
At the centre of the meeting is the future of the AI Act, the first attempt at regulating the technology in the world, as this is already under review via the Digital Omnibus. This single proposal amending the GDPR, the AI Act and the ePrivacy framework intends to rewrite their rules in the name of competitiveness.
Our Deregulation Monitor tracks the same pattern across digital policy, environmental protection, workers’ rights and corporate accountability. The stated target is administrative burden but the real impact is to lower standards when it comes to the use of your personal data.
Research by Corporate Europe Observatory and LobbyControl, traced eight major changes in the Commission’s proposals and found that seven closely mirror long-standing demands from Big Tech and its lobby groups. Public consultations and impact assessments were skipped, and someone filled the gap.
Where things stand
The AI side of the package is the most advanced. Parliament adopted the Omnibus in June 2026, and its headline change is time. Obligations for certain high-risk AI systems are postponed again this time from 2026 to 2027, others to 2028, and some transparency requirements are delayed too. That makes some sense, since European standards are incomplete and few bodies can assess conformity. But delay does not make the rules any simpler, and complex rules favour those who can afford lawyers, which start-ups and civil society organisations cannot. The package also introduced some improvements, including a new ban on AI systems designed to generate non-consensual intimate imagery and an EU-level sandbox with priority access for SMEs.
The data protection side is where the fight remains alive. In June, we joined more than 20 organisations in an open letter to Member States warning against narrowing the definition of personal data, widening AI-related processing and blurring the line between ePrivacy and the GDPR.
Then came a leak. As reported by The Brussels Times and summarised in our latest piece, a draft Irish presidency compromise would let companies use personal data to train and operate AI on the basis of “legitimate interest”, potentially without consent. It would also reintroduce an AI provision that Member States had removed. The Irish presidency rejects any suggestion that it is not impartial.
The rules we already have are not in great shape either. In July, we asked why the Commission applauds itself for enforcing parts of the Digital Services Act against a handful of very large platforms, while others, such as the right to a non-profiling recommender option, stay dormant.
Dublin, by invitation only
Ireland chairs the final stages of the Omnibus negotiations as holder of the Council presidency. Today, the Irish Council Presidency hosts the International AI Summit at the Royal Dublin Society. It is reported that about 1,500 people are expected, the guest list is not public, and at least three European Commissioners and four Irish ministers will attend.
The agenda is easy to summarise. 82% of the people going are executives of companies or business groups. The rest go to Commissioners, the Taoiseach, ministers and a single trade unionist. No civil society organisations, no academics. The panel on AI in healthcare features three pharmaceutical companies and an Irish AI start-up, but no health authorities or patient representatives. Presumably they will be told how it went.
The trade union slot was itself a late addition, made after Irish unions complained in July. It is a ten-minute session shared with an employers’ group. Civil society and media organisations have reportedly struggled to get on the guest list.
The setting makes sense once you look at the economics. Microsoft and Apple alone accounted for about 40% of Ireland’s corporate tax receipts last year. Ireland is not the only place where this dynamic exists, but it is a particularly vivid example of why the question of who sits at the table is not abstract.
The counter programme
As a response to being excluded from the AI Summit, civil society has organised its own programme. The Workers’ AI Counter Summit, a two-day event run by the Communications Workers’ Union, took place over the weekend, which aimed to reflect how AI is actually experienced by workers and communities. Along the summit, the Centre for Digital Ethics is also hosting a public-interest conversation alongside the summit, focused on the human values and institutions that technology should serve.
Who writes the rules
None of this requires assuming bad faith. As we put it, the question is whether the institutional circumstances create a credible risk that private interests exert disproportionate influence over public decisions. As Prof. Alberto Alemanno’s response to Draghi frames the broader point: faster AI adoption is not a strategy for shared prosperity unless we ask who benefits, who bears the costs and whose interests shape the rules.
Europe clearly wants AI sovereignty. The Commission’s call for seven AI gigafactories is open until 12 November. But sovereignty is not only about who owns the servers. It is also about who is in the room when the rules are written.
A summit can showcase the people who build AI. A rulebook should also hear from the people it is used on. The Omnibus is not final, and the guest list is not the policy. It does, however, tell you whose calls get returned.