Luca Bertuzzi

Austria has kicked off efforts to define what “digital sovereignty” means for Europe with a first draft of an EU declaration laying out guiding principles to shape future priorities, stressing that digital sovereignty is about autonomy, not isolation.
www.mlex.com/mlex/technol...

Austria has kicked off efforts to define what “digital sovereignty” means for Europe with a first draft of an EU declaration laying out guiding principles to shape future priorities, stressing that digital sovereignty is about autonomy, not isolation.
www.mlex.com/mlex/technol...

More broadly, the reports hint at a paradigm shift:
from measuring capacity to evaluating a model’s propensity to cause harm. They also outline a graduated compliance system. If confirmed, these approaches could reshape how GPAI obligations are applied under the AI Act. 4/4

Beyond the much-debated compute threshold, the researchers identify three other factors to classify GPAI models with “systemic risk”: safety benchmarks, reach & high-impact capabilities. They suggest regulators should set clear thresholds in these areas. 3/4

The studies go deep into how to technically define and assess GPAI models. One proposal borrows from cognitive psychology to determine whether an AI system is “general-purpose.” Another offers technical criteria to decide when a model is so altered it becomes a “new” one. 2/4

Last week, the EU's Joint Research Centre released a set of external studies on general-purpose AI models. They don’t officially represent the Commission’s views, but they could shape how the AI Act’s GPAI rules are interpreted in practice. 1/4👇
www.mlex.com/mlex/article...

No common specs in sight for the AI Act. The commission told MEPs today it isn’t planning to develop mandatory “common specifications,” even though the technical standards for the high-risk rules will be late for the Aug. 2026 deadline.
www.mlex.com/mlex/artific...

The European Commission has given its first preview of how the digital omnibus will amend the EU AI Act. Topics “under consideration”: interplay with other laws, governance, innovation, SME exemptions, and a possible timeline alignment.
www.mlex.com/mlex/article...

The European AI Board could be in for a major shake-up, as Austria is pushing to downsize the role of the Commission's AI Office in favor of an elected chair and clarify the role of the body's subgroups. Full story on @mlexclusive.
www.mlex.com/mlex/artific...

The Netherlands weighs in on the EU’s digital simplification agenda. Key message: focus on legal clarity, not deregulation. It also calls for streamlined governance of EU boards and, on the AI Act, urges simplification over delays.
www.mlex.com/mlex/article...

The EU Commission plans to publish its AI Act interplay guidelines, covering GDPR, product safety, platform & copyright laws, only in Q3 2026, right before or after the high-risk rules take effect.
www.mlex.com/mlex/article...

Digital sovereignty is on the agenda of the next European Council summit. Draft conclusions seen by MLex show EU leaders set to reaffirm the bloc’s regulatory autonomy and urge lawmakers to fast-track files related to simplification.
www.mlex.com/mlex/article...

The EU Commission has again delayed its decision on whether to propose a pause for the AI Act’s high-risk rules. An announcement was considered alongside the Apply AI Strategy, but dropped after the digital omnibus was moved up. Full story on MLex.
www.mlex.com/mlex/article...

What to expect from the EU’s upcoming Apply AI Strategy? The plan aims to build a sovereign AI ecosystem by boosting AI use across 10 key sectors. Initiatives include a study on genAI legal challenges, open model competitions, & research on AI agents.
www.mlex.com/mlex/article...

EU countries are aligning on cloud sovereignty, with most member states backing the direction the Commission has taken with the Cloud & AI Development Act. The Commission eyes finalizing EUCS certification & embedding it in procurement rules.
www.mlex.com/mlex/article...

Ombudswoman opens inquiry concerning the development of EU standards for artificial intelligence European Ombudswoman Teresa Anjinho has opened an inquiry into how the European Commission ensures transparency, inclusiveness, and accountability in the ...

The EU Ombudswoman has opened an inquiry into the AI Act’s standardization process because CEN/CENELEC are not required to disclose the names of participants or publish meeting minutes. The question is whether this could create further delays.
www.ombudsman.europa.eu/en/news-docu...

EU countries split on the AI Act pause. Many back a delay, others warn it’s counterproductive, & heavyweights remain noncommittal. The Commission will decide soon, with only “quick fixes” planned in the digital omnibus & a broader Fitness Check to follow.
www.mlex.com/mlex/article...

The European AI Board is set to change its leadership model, currently tied to the Council’s rotating presidency, to an election-based system like the @EU_EDPB. First talks on the new rules start in October, with the shift expected next year.
www.mlex.com/mlex/artific...

The Commission has just published a draft guidance and a template for serious incident reporting under the EU AI Act, together with a public consultation open until November 7.
digital-strategy.ec.europa.eu/en/consultat...

The EU’s ICT Supply Chain Security Toolbox is almost ready. It will cover all digital technologies, from connected cars to AI models. Adoption expected by November, with countries asked to exclude high-risk vendors & add cybersecurity rules in procurement.
www.mlex.com/mlex/article...

In July, Sweden's PM sent a letter to the Commission President with a wishlist for the upcoming digital omnibus, including a grace period for the AI Act, guidance on more uniform enforcement, and a reform of the telecom sector. Full letter on MLex.
www.mlex.com/mlex/article...

Is AI vendor lock-in a sovereignty issue? Denmark and other Nordics are pushing the debate, and the Commission’s consultants are similarly exploring pro-competition measures for the upcoming Cloud & AI Development Act.
www.mlex.com/mlex/article...

The Commission will split the AI Act guidelines on high-risk systems. Classification guidance is due by Feb. 2, 2026, while separate rules on obligations, substantial modification and the AI value chain are planned for Q2–Q3 2026. Details on @mlexclusive.
www.mlex.com/mlex/article...

AI is taking center stage in the DMA's first review. In my latest deep dive for MLex, I describe how the Commission is framing the issue, the many legal questions that it is facing, and the three policy options that are currently on the table.
www.mlex.com/mlex/article...

The EU’s long-awaited ICT Supply Chain Toolbox will be discussed on Sept. 25. It targets high-risk foreign suppliers, like Huawei & ZTE. Now linked to the Cybersecurity Act review, it could shift the EU's approach from soft-law guidance to binding rules.
www.mlex.com/mlex/article...

The 🇪🇺 Commission just opened a public consultation confirming the digital omnibus will cover the AI Act. It will propose “targeted adjustments” to ensure effective, predictable enforcement. Feedback is open for 4 weeks.
ec.europa.eu/info/law/bet...

Poland is pushing to delay the EU AI Act’s penalty regime for high-risk AI by 6–12 months. In a non-paper, Warsaw calls for suspending Article 99 sanctions, clearer guidance, alternatives to technical standards, and wider use of AI sandboxes.
www.mlex.com/mlex/artific...

The Commission is gearing up to launch the EU code of practice on generative AI transparency in early Nov. Meanwhile, it has asked member states to back the creation of a new sub-group within the AI Board to shape the code and accompanying guidelines.
www.mlex.com/mlex/article...

The Commission is convening five “reality check” discussions to help shape its upcoming digital omnibus proposal. These invite-only focus groups will assess how EU rules on AI, cyber, data, privacy, and digital identity work in practice.
www.mlex.com/mlex/article...

The EU Cloud & AI Development Act is delayed to at least March 2026, pending political direction on cloud sovereignty. Meanwhile, the idea of “effective European control” is being considered as benchmark for determining whether a cloud provider can be “trusted.”
www.mlex.com/mlex/article...