🤖 EU AI & Sovereignty Weekly
August 4, 2026
This week’s EU AI & Sovereignty update. The AI Act enforcement started Sunday, August 2 — a historic milestone that will dominate the agenda this autumn. This edition focuses on the new European Commission publications not yet covered in this newsletter: the enforcement itself, the AI Gigafactories call, the Code of Practice on AI transparency, the transparency guidelines, the entry into force of the AI Omnibus and the fourth GPAI taskforce meeting.
wetgeving
🔥 HOOG
Commission starts enforcing AI Act and new transparency rules on August 2
From August 2, 2026 the AI Office and national authorities began enforcing the AI Act. New transparency rules require chatbots to tell users they are talking to AI, deepfakes to be labelled, and AI-generated content to carry machine-readable markings. This is the moment the AI Act stops being a text and becomes enforced reality — expect the first concrete actions in the coming weeks.
investeringen
🔥 HOOG
EU launches AI Gigafactories call — up to €30 billion investment
The EU has launched a call to set up up to seven AI Gigafactories across Europe, a cornerstone of its ambition to become the ‘AI Continent’. Industry-led and backed by up to €10 billion in EU and national funds, it aims to unlock at least €20 billion in private investment and provide access to infrastructure for training, inference and fine-tuning of advanced frontier models. Together with the 19 AI Factories, this is Europe’s bet on building its own compute capacity rather than renting it from elsewhere.
wetgeving
📌 MEDIUM
Over 190 organisations sign Code of Practice on transparency of AI-generated content
Around 190 organisations signed the Code of Practice on Transparency of AI-generated Content, just before the AI Act marking requirements took effect on August 2. It gives providers and deployers of generative AI a streamlined, predictable and legally certain route to comply. Signatories include Google, Meta, Microsoft, OpenAI, Anthropic, Mistral, Aleph Alpha and Synthesia; about half are small and young companies. Two taskforces launch in September.
wetgeving
📌 MEDIUM
Commission publishes guidelines on transparency obligations for AI systems
The Commission published guidelines helping providers and deployers meet the transparency obligations of Article 50 of the AI Act, applicable since August 2. Providers must design AI systems so users know they are interacting with AI, and label AI-generated or manipulated content with machine-readable markings. Deployers must also inform people when exposed to deepfakes and emotion recognition or biometric categorisation systems.
wetgeving
📌 MEDIUM
AI Omnibus enters into force — simplification with longer timelines
On July 27 the AI Omnibus entered into force EU-wide, extending timelines and simplifying the AI legal framework while keeping strong safety and fundamental-rights safeguards. It lightens the compliance burden for smaller businesses and expands regulatory sandboxes, including an EU-level one. Key dates: high-risk AI in Annex III applies from December 2, 2027, and high-risk AI embedded in physical products from August 2, 2028. A ban on ‘nudification’ apps is also introduced.
Commissie
📌 MEDIUM
Fourth GPAI taskforce meeting on safety, security and copyright
The fourth Signatory Taskforce meeting under the GPAI Code of Practice focused on the Safety and Security chapter and the Copyright chapter. The AI Office explained how model usage can be part of post-market monitoring and systemic risk assessment, and discussed ‘marginal-risk’ clauses in safety frameworks — which should only be invoked in exceptional circumstances with appropriate safeguards. On copyright, providers must give rightsholders information about the web crawlers and robots.txt settings they use, made public.
wetgeving
📌 MEDIUM
New guidance on implementing the Cyber Resilience Act — AI and cybersecurity
The Commission published guidance to help businesses apply the Cyber Resilience Act (CRA), clarifying which products fall under the law, what counts as a ‘substantial modification’, and how to meet reporting and risk-assessment obligations, with extra attention for micro and small enterprises. It stressed that recent developments in frontier AI models with cyber capabilities make swift and correct implementation all the more urgent. Reporting obligations start September 11, 2026.
Commissie
🔥 HOOG
Commission fines Google €890 million for DMA breaches
The Commission took two non-compliance decisions against Google for breaches of the Digital Markets Act: self-preferencing in Google Search (€460 million) and anti-steering restrictions on Google Play (€430 million). Google gave its own services — shopping, hotels, transport and sport — more prominent placement than third parties and limited app developers from directing customers to cheaper alternatives. One of the largest DMA fines to date, and Google must end the practices.
technology
📌 MEDIUM
Commission: TikTok breaches DSA — unsafe accounts for minors
The Commission sent TikTok preliminary findings that minors’ accounts do not meet the safety standards of the Digital Services Act. Minors can set their account to ‘public’, making their content visible to everyone, and content from 16-17 year-olds is recommended to random users via the For You feed. TikTok must change the default settings so minors’ content is only visible to accepted users by default and not recommended via the For You feed.
technology
📌 MEDIUM
New GenAI pilots for public administrations get underway
Three pilot projects supporting generative AI in public services — FLOODS & DROUGHTS, EUNOMIA.AI and EuropAI — signed their grant agreements on July 1. Selected under the ‘GenAI for the public administrations’ call of the Digital Europe Programme, they test trustworthy European GenAI solutions for concrete government needs. They feed the Apply AI Strategy and should deliver scalable, interoperable solutions other European public services can adopt — a concrete step towards sovereign AI in the public sector.
📈 Key Trends
- • AI Act enforcement is now real — the first complaint and whistleblower tools are operational, and concrete enforcement actions are expected in the coming weeks.
- • The AI Gigafactories call (up to seven factories) is Europe’s most important sovereignty dossier — a direct bid to build its own compute and cut dependence on foreign infrastructure.
- • Transparency is the unifying theme: labels, deepfake markers, machine-readable markings and the Code of Practice shape how generative AI will be held to account.
- • The DMA and DSA are being actively enforced — the €890 million Google fine and the TikTok preliminary findings show platform oversight is no longer theoretical.
- • The AI Omnibus eases the compliance burden for smaller businesses while keeping safety safeguards — a balance that will define the next phase of EU AI law.
Wat vond je van dit bericht?
