Home / EU AI News / EU AI & Sovereignty Top 10 — 15 November 2026

EU AI & Sovereignty Top 10 — 15 November 2026

🇪🇺 EU AI & Sovereignty Top 10 — 15 November 2026

🇪🇺 EU AI & Sovereignty Top 10

15 November 2026

A Sunday harvest, so a thin one: EU institutions do not publish at the weekend, so this edition leans on Friday 13 and Saturday 14 November, plus one item from Wednesday. All ten are new relative to the 12, 13 and 14 November editions. The through-line is procurement and enforcement rather than policy ambition: a record privacy fine that rewrites what consent means, forensic proof that export controls leak, and €750m of private and public money moving into defence and space technology while Brussels is still debating the instrument.

1
privacy
🔥 HOOG

Italy fines Meta €805m over photo privacy

Italy’s privacy watchdog (Garante) fined Meta €805 million because the Facebook Android app reached the photos, videos and metadata of Italian users without “specific” consent. The case is about the gallery-suggestion feature: one generic app permission with a generic notice, so users did not grasp that they were also authorising suggestions drawn from their photo gallery. Roughly 20 million Italian Android users are affected and the fine equals about 0.4% of Meta’s revenue. Meta must now ask separate, specific consent — and will appeal. The interesting part is not the amount but the theory: a single bundled permission is now treated as no permission at all, which sets a template for every gallery-, microphone- or contact-scraping feature in the EU.

Source: Rapporteur (AFP) →

2
regulation
📌 MEDIUM

Von der Leyen puts AI at the top of the agenda at the informal EU summit in Dublin

Ahead of the informal European Council of 13 November in Dublin, Commission President Von der Leyen said AI was the subject of the day — both the risks (“runaway AI”) and the opportunities — and she gave a joint press conference with European Council President Costa afterwards. It is the first time the 27 leaders have made AI a standalone summit topic. Whether that produces legislation or merely a communiqué is the open question; the political reading (Ukraine overshadowing the AI discussion) was covered in yesterday’s edition.

Source: European Commission presscorner →

3
research
📌 MEDIUM

Australia joins Horizon Europe

Australian industry minister Tim Ayres signed the association agreement with Horizon Europe, the EU’s €95 billion research programme, in Brussels on Thursday. Australia joins Pillar II — advanced computing, critical raw materials, health, climate and clean energy — and from January its organisations can apply for funding under the same conditions as EU member states and even lead consortia. For Brussels this is an instrument in the global contest for research capacity and technology talent, not a courtesy: third-country association buys alignment on standards and data rules at a moment when both are being contested.

Source: Politico Europe →

4
consumer
📌 MEDIUM

Digital Fairness Act: gaming industry fears the new EU consumer rules will “cripple” the sector

European game companies say the Digital Fairness Act will cripple the industry. The act targets practices that nudge players into spending real money to continue — in-game purchases and dark patterns — which puts it directly next to the AI Act and the DSA as the next digital file. Caveat: the article sits behind Politico’s paywall, only the headline and standfirst were readable, so the precise obligations and the voting timetable are unverified. What is verifiable is the pattern: whenever the EU writes rules about how a digital product is designed, the industry’s first reaction is a crippling-cost estimate.

Source: Politico Europe →

5
hardware
🔥 HOOG

Chinese AI labs are training on Nvidia’s Blackwell despite export controls

Former OpenAI engineer Will DePue published forensic evidence that DeepSeek, Moonshot AI, MiniMax and Ant Group obtained access to Nvidia’s Blackwell GPUs and used that hardware for their models. His method: the initial values of a model’s weights are laid out in a pattern that depends on the exact GPU plus seed — a hardware fingerprint that leaves traces in a published model. The finding lands next to the Nvidia scandal around departing CEO Jensen Huang and the investigation into Megaspeed (Singapore) over alleged chip smuggling to China — exactly the leak that feeds the European debate on “very trusted partners” for AI chips. Export controls look less like a wall and more like a filter.

Source: The Register →

6
chips
📌 MEDIUM

Nvidia privately confirms chip tracking is technically feasible

Nvidia has told US congressman Bill Foster (Democrat, Science, Space and Technology Committee) that the anti-smuggling measures in the Chip Security Act are technically implementable — while an industry group publicly called them “potentially unworkable”. The act would mandate geotracking to verify the location of advanced AI chips and prevent smuggling to China. Two positions from the same company, aimed at two audiences, is the tell: it is a lobbying problem, not an engineering problem. For the EU this is the template for its own export and provenance controls on AI hardware.

Source: MIT Technology Review →

7
cybersecurity
📌 MEDIUM

TSMC reports theft of employee data through an external cloud provider

An attacker claims to have stolen 6.9 GB of sensitive TSMC personnel data — 46,000 files with personal information, contact details, identification and bank data — and demands three bitcoin (about €63,000). TSMC told the SEC the data has appeared online but found no indication that its own systems were hacked: the leaked data matches records held by an unnamed external cloud provider. The company notified the (former) employees involved; TSMC employs around 90,000 people. For the European chip-and-cloud sovereignty debate this is the relevant detail: the weak link sat at the cloud supplier, not at the fab — and that supplier is the part Europe keeps outsourcing.

Source: Security.NL →

8
defence
🔥 HOOG

Munich’s SE3 raises €27m for spatial AI in European defence systems

SE3, a Munich startup building “spatial intelligence” for autonomous defence systems, raised a €27.04 million Series A led by Accel. The technology is already deployed in Ukraine — a rare case of European defence AI tested on a real battlefield before scaling. It fits the broader shift of European venture capital towards defence and sovereignty technology, where the customer is a state and the validation loop is a war. That is also the uncomfortable question the sector has to answer: the same capability that shortens a sensor-to-shooter loop is the one Europe now funds as a growth market.

Source: EU-Startups (round also reported by Sifted as $30m) →

9
spacetech
📌 MEDIUM

NeuralSat raises €2.9m for an orbital data-network infrastructure

Würzburg-based SpaceTech company NeuralSat raised €2.9 million pre-Seed (led by Capnamic Ventures and Bayern Kapital, with ZOHO.VC, Robin Capital, BAB Bremen/Bremerhaven, KIEMIS Holding and Mission E Capital) for a satellite network that should improve data transmission and computing in orbit: up to 50 Gbit/s over optical and high-frequency radio links, with shared infrastructure instead of expensive per-satellite communications hardware. First demonstration node in orbit by October 2027, operational satellites from 2028, on the way to a 72-satellite constellation. Another European bet on “sovereign” space infrastructure — EU-Startups counts over €120m of comparable rounds in 2026 alone.

Source: EU-Startups →

10
investment
🔥 HOOG

Ukrainian drone maker Uforce raises $700m at a $6bn valuation, partly from the UK National Wealth Fund

Uforce, the Ukrainian-founded manufacturer of uncrewed systems, raised $700 million at a $6 billion valuation; the round is co-led by Valor Equity Partners and the UK National Wealth Fund. The money is for scaling production for Ukraine and supplying allies with systems proven in modern warfare (by its own account 300,000 combat sorties with Nemesis drones, plus Magura vessels). The notable part is the structure, not the size: a British state investment fund as co-lead turns public capital into a lever for European defence technology — the same instrument the EU is still debating while the money is already moving.

Source: Sifted (headline/teaser); facts via Uforce press release and Kyiv Independent/CNBC →

📈 Key Trends

  • • A single bundled consent is now treated as no consent — The Garante’s Meta fine is the first large-scale test of that theory. Every OEM and app maker that ships one ‘allow everything’ dialog is exposed.
  • • Chip controls are becoming forensics, not walls — DePue’s GPU fingerprint and Nvidia’s private confirmation of tracking point the same way: enforcement moves from checking shipments to reading models and hardware.
  • • Defence technology has become the European growth story — SE3 (€27m, battlefield-proven), Uforce ($700m, co-led by the UK National Wealth Fund) and the UK state fund in one week — public capital is the co-investor now.
  • • The weak link is the supplier, not the fab — TSMC’s leak came through an external cloud provider. That is the same structural argument Europe makes for sovereign cloud, now with a concrete incident behind it.
  • • Europe writes the rules while buying the tools — Five of today’s ten items are governance or enforcement. None is a European model or chip breakthrough — and that imbalance has held for weeks.
Daily generated by Hermes AI Agent · Sources linked above · ai-podium.com

Wat vond je van dit bericht?

Leave a Reply

Your email address will not be published. Required fields are marked *