29 juli 2026 · 13:23
EU AI Act Transparency Live: Labels, IonQ & 6G
EU AI Act transparency rules come into force this Sunday, requiring clear labelling of every AI-generated image, video and text on platforms serving European users, and the compliance burden lands on every business running generative AI in its workflow. In the same episode we cover IonQ winning regulatory approval to acquire US foundry SkyWater, the first fully vertically integrated quantum company, alongside Fitch's warning that the AI boom now represents a top global credit risk. The State of Tech, The European Edition walks you through what each development means for European professionals, investors and policymakers.
Beluister deze aflevering:
Transcript
Samantha: Welcome to The State of Tech, The European Edition, Wednesday July twenty-nine, 2026. I'm Samantha Lawrence.
Bob: And I'm Bob Russell. Today: Europe's AI transparency rules go live this weekend, IonQ gets the green light to swallow a US chip foundry, Fitch sounds the alarm on an AI market correction, the US pulls together a global coalition for the next generation of mobile networks, and to close out, two things you can actually use today: the free European writing app Novelist, and a handy browser extension called Consent-O-Matic. Let's start with Brussels.
Europe's AI transparency rules go live on Sunday, forcing companies to label every AI-generated image, video and text.
Samantha: This is the one European users will actually notice. From this Sunday, the transparency chapter of the EU's AI Act starts biting. Anything generated or meaningfully altered by AI, so deepfakes, synthetic images, cloned voices, machine-written text, has to be clearly labelled so a regular person can tell it apart from the real thing. That means visible marks on images and video, disclosures on chatbot conversations, and machine-readable signals baked into the files themselves.
Bob: And the scope is broad. It's not just the big American platforms. Any company offering AI systems to European users falls under it, from a small marketing agency using a text generator to the giants running foundation models. Providers have to build the labels in, and deployers, the businesses actually using the tools, have to make sure the disclosure reaches the end user. Fines can run into the millions or a percentage of global turnover, whichever hurts more.
Samantha: The stated goal is trust. Brussels has been watching the deepfake problem get worse through two election cycles, and the political mood shifted from "let's wait and see" to "let's make people label this stuff." Whether the labels actually help is the open question. Research on misinformation warnings is mixed at best. A watermark only works if people notice it and if the platforms preserve it when content gets re-uploaded.
Bob: There's also the technical side. Watermarking AI text is genuinely hard. Strip out a few words, paraphrase, and the invisible signal can disappear. Image watermarks are more robust, but not bulletproof. So the law leans heavily on visible labels as the backstop, which puts the compliance burden squarely on the companies deploying these tools day to day. Expect a scramble this week from every European marketing department that quietly runs everything through a chatbot.
Samantha: For European citizens the practical effect is you'll start seeing more little "AI-generated" tags across your feeds, your news sites, your customer service chats. For companies it means auditing every workflow that touches generative AI. And for the rest of the world, Brussels is once again writing the default rulebook. If a US or Asian company wants to serve European users, they build to the AI Act. The Brussels Effect, round three, after data protection and platform rules.
IonQ gets the final regulatory nod to buy US chip foundry SkyWater, building a full-stack quantum supply chain.
Bob: Staying with big structural moves, quantum computing company IonQ has cleared the last regulatory hurdle to buy SkyWater Technology, the largest chip foundry that operates exclusively on US soil. The deal makes IonQ the only quantum company that owns its own manufacturing base from end to end, from the physics research through to the silicon that runs the control systems. SkyWater keeps operating as a subsidiary and keeps serving its other customers.
Samantha: The strategic logic is straightforward. Quantum computers still need a lot of classical chips around them, custom control electronics, specialised photonics, packaging that nobody else makes in volume. If you're waiting in line at a commercial foundry alongside everyone building smartphone chips, you're stuck. Owning the foundry means IonQ can prioritise its own roadmap and iterate hardware faster. It also insulates them from the export control turbulence that keeps hitting the semiconductor world.
Bob: For Europe this is a signal worth reading carefully. The continent has quantum ambitions, real ones, with programmes in France, Germany and the Netherlands, and companies like IQM, Pasqal and Quantinuum's European operations. But none of them own their own foundry. If IonQ's bet pays off and vertical integration becomes the winning model, European quantum firms will need to answer how they secure their own chip supply. That question gets sharper when the leading US player owns its factory outright.
Samantha: And there's a broader industrial policy point. The US has been quietly consolidating quantum, AI and chips into a single strategic bucket. This deal fits that pattern. Europe's response so far has been fragmented, with national champions rather than continental ones. Industry groups have long argued that without a joined-up approach, Europe risks becoming a customer of American quantum infrastructure rather than a builder of it.
Samantha: Quick interruption. If you listen to The State of Tech regularly, hit that like button and subscribe, that way you'll never miss an episode. Okay, moving on.
Fitch warns the AI boom is now a major global credit risk as valuations outrun returns.
Bob: Ratings agency Fitch has put a very blunt line in its third-quarter Global Risk Outlook: the artificial intelligence boom, and the potential for a sharp correction, is now one of the top global credit risks. Their argument is that AI spending has scaled faster than the revenue to justify it, tech valuations are stretched, and the sheer size of the capital going into AI infrastructure means a correction would ripple across the whole economy, not just tech stocks.
Samantha: Fitch specifically flagged Asia, where AI-linked stocks have already started sliding. That's notable because Asia has been the momentum engine for the trade over the past year, driven by memory makers, packaging firms and equipment suppliers riding the data centre wave. When the momentum crowd starts stepping back there, it usually reaches Europe and the US within weeks. And the credit angle matters because a lot of AI infrastructure has been financed with debt, huge multi-year commitments for chips, power and buildings.
Bob: The scale is the thing Fitch keeps returning to. When individual companies are committing tens of billions to single data centre projects, and vendors are extending guarantees to underwrite each other's spending, the interconnections start to look like a web. If one node wobbles, the exposure spreads. Fitch is not calling a crash, they're calling a systemic risk that would deserve serious attention if returns disappoint.
Samantha: European pension funds, insurers and banks have absorbed a lot of this exposure indirectly, through index funds heavy in US tech, through corporate bonds from hyperscalers, through infrastructure debt tied to data centre projects. Fitch's warning is essentially a nudge to European financial institutions to check their concentration risk. It also lands at a politically awkward moment, with Brussels still pushing for more AI investment as an industrial priority.
The US launches a global 6G coalition with more than twenty governments, including Europe, to set standards for the next mobile generation.
Bob: Moving to telecoms, and this is our non-AI story of the day. The US National Telecommunications and Information Administration has launched what it calls a Call to Action for 6G Leadership and Security. More than twenty partner governments have signed up, including European countries and Indo-Pacific allies. The aim is to line up a shared vision for what comes after 5G, before the standards battles really kick off.
Samantha: The lessons from 5G are shaping this. That rollout got tangled in geopolitics, security concerns around specific vendors, and a fragmented approach between allies that let one or two suppliers dominate. This time Washington wants the standards conversation started early, with a strong emphasis on security, interoperability between different vendors, and open network architectures so no single company can lock the ecosystem.
Bob: For Europe there's a real balancing act here. European telecom equipment makers like Ericsson and Nokia have a genuine seat at the table, arguably a bigger seat than they had in the 5G era. But European governments also want to preserve their own industrial policy space, particularly on which vendors can participate in critical infrastructure. Joining the US coalition helps set common rules, but Brussels will not want to be dictated to on procurement.
Samantha: The commercial 6G timelines still point to around 2030, so this is early-stage diplomacy rather than immediate technology. But the standards decisions get locked in years before deployment. Whoever shapes the specifications shapes the market. And there's a deliberate line in the announcement about trusted AI being integral to how 6G networks are built and operated, which tells you where the security thinking is heading.
Bob: The bottom line is that mobile networks are becoming as much a geopolitical question as an engineering one. European operators and their customers should watch which vendors and which countries end up inside the trusted circle.
The free European writing app Novelist helps you plan and finish a book without any subscription.
Samantha: To close out, two things you can actually use today. First up, a writing app called Novelist. It's aimed at anyone who has been meaning to write a book, a long report, or a serious blog series but keeps getting lost in the structure. Novelist runs on Windows, Mac, Linux and Android, it costs nothing, and it does one thing well: it lets you plan chapters, sketch characters, keep timelines straight, and actually write the prose in the same place.
Bob: What makes it stand out from the usual word processors is how it separates the messy planning brain from the writing brain. You get a corkboard view for outlining scenes, a character sheet so you stop forgetting your protagonist's eye colour halfway through, and a distraction-free writing mode when you're ready to just type. Users often mention that it feels less intimidating than the professional screenwriting tools, but more structured than a blank document that scrolls forever.
Samantha: The privacy angle is worth noting for European listeners. Novelist stores your project files locally by default, so your half-finished novel isn't sitting on somebody else's server being scanned for training data. You can sync to your own cloud folder if you want, but you don't have to. And because it's not subscription-based, you're not paying a monthly fee to keep access to your own manuscript, which has become a real complaint about some of the bigger writing platforms.
Bob: The typical time investment is about twenty minutes to set up a project, import any existing chapters, and get the outline going. From there it's really just about opening the app and adding a few hundred words at a time. It's the kind of tool that removes the excuse of "I don't have the right software yet." Because now you do, for free.
Consent-O-Matic is the free browser extension that clicks away every cookie banner for you.
Samantha: Second thing to try today, a browser extension called Consent-O-Matic. If you're European, you know the pain. Every website greets you with a cookie banner. Some are honest, most are designed to make "reject all" as hard as possible, hiding it behind three menus and a switch called "legitimate interest" that isn't really either. Consent-O-Matic just handles all of that for you in the background.
Bob: The extension was built by researchers at Aarhus University in Denmark, and it's free and open source. It works on Chrome, Firefox, Edge and Safari. Once installed, it detects the cookie consent pop-ups automatically and clicks through them according to your preferences. You set your privacy choices once, do you want analytics cookies, marketing cookies, personalisation cookies, and from then on the extension enforces those choices on every site you visit.
Samantha: What makes it particularly useful in Europe is that it targets the specific consent frameworks that dominate here, the ones the big cookie banner vendors use. The extension recognises them, opens the hidden settings pages, and applies your preferences the way the law says the sites should have offered in the first place. Users report it saves them dozens of clicks a day and, more importantly, actually rejects the tracking they thought they were rejecting.
Bob: There's a research angle too. The team behind it has published papers documenting how deceptive many cookie banners are, and the extension is partly a demonstration that a compliant experience is technically possible, if the sites wanted to build one. Installation takes about a minute from your browser's extension store. After that, the internet just gets quieter.
Samantha: Today we covered: Europe's AI labelling rules going live this weekend, IonQ buying US chip foundry SkyWater, Fitch warning on AI as a credit risk, the US launching a global 6G coalition, and two things to try yourself: the free writing app Novelist and the cookie banner extension Consent-O-Matic.
Bob: Want to know more or react? Visit stateoftech.eu or email us at info@doorzetters.net.
Bob: State of Tech, the tech world in 15 minutes.