US AI air defence in Europe, Korea chip exports up 259%

Samantha: Welcome to The State of Tech, The European Edition, Sunday November one, 2026. I'm Samantha Lawrence.

Bob: And I'm Bob Russell. Today: US forces in Europe switch on AI-powered defences against missile and drone swarms, South Korea's chip exports more than triple, Meta reportedly builds a humanoid robot with a Chinese startup, and attackers exploit flaws in GitLab and Cisco firewalls. Plus two things to try yourself: Microsoft PowerToys and NotebookLM. Let's start with Europe's skies.

US forces in Europe deploy AI systems built to fight off mass missile and drone attacks.

Samantha: American forces stationed in Europe have started deploying AI-powered defence systems built for a nightmare scenario: hundreds of missiles and drones arriving at the same time. It's the result of months of testing and integration throughout 2026. The systems fuse data from radars, aerial surveillance and satellites in real time, then predict where every incoming object is heading. The point is speed: sorting hundreds of targets in the seconds an attack actually leaves.

Bob: Then the software takes over the triage. According to the reports, it ranks threats automatically and coordinates the response, allocating defensive weapons against hundreds of simultaneous aerial targets. In practice that means deciding which drone gets which interceptor, and in what order. That matters because drone swarms are designed to overwhelm defences through sheer numbers. Many machines, all at once. This deployment is the US military's answer to that tactic, moved out of the test range and into the field.

Samantha: It's a real shift in how armies prepare for modern warfare. Air defence used to be about a handful of high-value threats, a fighter jet or a single cruise missile. Mass swarms flip that logic, and automated prioritisation tries to make sure the right threats are handled first. What the reports don't spell out is how much of the final decision stays with human operators. That question will follow this technology wherever it goes.

Bob: And this lands directly in Europe. These systems sit on European soil, protect European airspace, and strengthen NATO's collective defence against exactly the kind of sophisticated aerial threats the alliance is preparing for. For European capitals, that's a concrete security gain. At the same time, it means part of Europe's air protection now runs on American AI software. That raises a practical question for European governments: build similar capabilities at home, buy them, or plug into what the US already fields.

Samantha: The bottom line: AI in defence is moving from testing into live deployment, right where people live. For European governments, it adds weight to two discussions at once: how to fund modern air defence, and how much control humans keep over automated weapons. And for listeners, the shift is clear. The next generation of air defence is less about one powerful missile and more about software fast enough to keep up with a swarm.

South Korea's October exports hit a record 115 billion dollars, powered by surging chip sales.

Bob: South Korea just posted the biggest export month in its history. Exports hit a record 115.3 billion dollars in October, up 93.7 percent on a year earlier. That's close to double. The engine is semiconductors: chip exports alone jumped 259 percent to 56.4 billion dollars. So roughly half of everything Korea shipped abroad last month was chips. The trade ministry credits two things: robust global demand for semiconductors, and sustained spending by big tech companies on AI infrastructure.

Samantha: Those figures show where AI money physically ends up. Every new AI data centre needs memory and processors, and this report shows how much of that demand is flowing to Korea. When tech giants announce another round of AI spending, part of that bill lands here, as chips in shipping containers. A 259 percent jump in a single year is not a gentle trend line. And notable is how concentrated Korea's export boom has become, with chips now carrying about half the load.

Bob: There's a tension with what we reported last time, when investors wiped trillions off big tech over fears that AI spending is outpacing visible returns. Whatever the stock market thinks, these export numbers say the spending itself hasn't slowed. Companies are still ordering chips at a record pace. For Europe, the surge is being read as a positive signal for European technology supply chains, because firms that feed into the chip industry stand to benefit when demand climbs this fast.

Samantha: And there's an investment angle. Numbers like these strengthen the case that the market for advanced semiconductors is large and still growing, which could encourage further investment in chip manufacturing and research inside Europe. The flip side is concentration: when half of a major exporter's shipments ride on AI demand, any slowdown in data centre building would be felt far beyond Korea. For now, though, the October figures point firmly in one direction: up.

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.

Meta has reportedly built a humanoid robot prototype called Max with Chinese startup Fourier.

Bob: Meta has reportedly built a humanoid robot. The prototype, internally called Max, was developed together with Chinese robotics startup Fourier. It weighs about 106 pounds, roughly 48 kilos, and according to the reports it was built in just three months. That is a short timeline for a machine this complex, and it shows how quickly humanoid hardware can now be put together when the right partners team up.

Samantha: And Meta isn't stopping at version one. A second prototype is already in development, with an improved torso and hands, a projector, and additional sensors to help it navigate. According to the reports, the bigger picture is household tasks. The work signals Meta's push to bring AI into everyday chores, and potentially to give its Muse AI agent a physical form. In other words, the assistant could one day move from software into a machine that walks around your home.

Bob: Meta's own line is more cautious. The company says this is internal research, not a product, so nobody should expect to order a Meta robot for Christmas. Still, research at this pace is a signal of how seriously Meta is exploring physical AI. And it's part of a wider trend of tech giants exploring advanced robotics for everyday use. Just last episode we covered Foxconn putting more than ten humanoid robots to work in its Texas facilities.

Samantha: The difference is the setting. A factory is a controlled space. A home is messy, private and full of people. A household robot packed with navigation sensors would, by design, be mapping the spaces people live in. Notable, too, is the partnership itself: a major American platform company working with a Chinese startup to build the hardware, and getting a prototype out of it in a single quarter.

Bob: For Europe, the reports point to two areas. European robotics research will be watching how fast a big platform company can move with the right hardware partner. And Europe's market for home AI assistants could shift if Meta ever brings this kind of device to consumers. If that happens, a robot moving through living rooms would land in a region with strict rules on personal data, and those rules would shape how it could be sold and used here.

US cyber agency flags actively exploited flaws in GitLab and Cisco firewalls, with a deadline today.

Samantha: Time to check your patches. The US Cybersecurity and Infrastructure Security Agency has added two flaws to its catalogue of vulnerabilities that attackers are actively exploiting. Not theoretical, already in use. The first is a maximum-severity flaw in GitLab, the platform many companies use to store, manage and build their software code. It allows remote code execution, which in plain terms means an attacker can run their own commands on the server from the outside.

Bob: The second is a high-severity flaw in Cisco Secure Firewall. A firewall is the gatekeeper that decides what traffic gets into a company network and what stays out. This bug allows a denial-of-service attack: in plain terms, an attacker can disrupt the firewall and knock the service offline. And here's the urgent part. American federal agencies were told to fix both problems by November first. That's today. Any agency that hasn't patched yet is now out of time.

Samantha: These two flaws hit different sensitive spots. A GitLab server can hold a company's source code, the instructions behind its apps and services. An attacker able to run commands on that server could potentially reach that code or tamper with it. A firewall going down is a different problem: the warning is about service disruptions, and a network without its gatekeeper working is harder to protect. Active exploitation is the reason these two landed on the urgent list.

Bob: For Europe, the deadline only binds American federal agencies, but attackers don't check passports. European organisations run GitLab and Cisco firewalls too. The practical step for anyone responsible for those systems: check which version you're running, install the available patches, and look through your logs for anything unusual. If you're not in IT yourself but your employer uses either product, a short message to your IT team asking whether this has been handled takes a minute.

Microsoft PowerToys adds a free toolbox of small time-savers to any Windows computer.

Samantha: To close out, two things you can actually use today. First up, Microsoft PowerToys. It's a free, open-source collection of small utilities from Microsoft for Windows 10 and 11, filling the little gaps Windows itself leaves. You install it from the Microsoft Store or from GitHub, which takes a few minutes, and then a single settings window lets you switch each tool on or off.

Bob: Users often single out the window layout tool. On a big monitor you can carve the screen into zones and snap apps into them, so your email, browser and documents each get a fixed spot. Another tool renames hundreds of files at once with search and replace, handy for holiday photos stuck with meaningless camera names. And there's an image resizer built into the right-click menu, so shrinking twenty photos for an email takes seconds instead of opening an editor.

Samantha: A few lesser-known tools cover other daily annoyances. One grabs text from anything on your screen, a scanned letter, a screenshot, a paused video, and turns it into text you can paste. Another pins any window so it always stays on top, handy for keeping a recipe or a video call visible while you work. There's a colour picker for anyone who designs, and a quick launcher you open with a keyboard shortcut to find apps and files without touching the mouse.

Bob: Now the limits. PowerToys is Windows only, so Mac and Linux users will have to look elsewhere. Some tools need administrator rights, which means on a work laptop your IT department may have to approve the install. And because it's a collection, it can feel like a lot at first. A simple start for this afternoon: install it, try the window layouts and the text grabber, and see whether they save you clicks by the end of the week.

Google's NotebookLM turns your own documents into an AI assistant that cites its sources.

Samantha: The second one is AI: NotebookLM from Google. The idea is simple. You upload your own material, PDFs, Google Docs, web pages or YouTube videos, and then ask questions about those sources. It works in a web browser and as a phone app, and you can start for free with a Google account. Think of it as a research assistant that focuses on your documents rather than a chatbot answering from the whole internet.

Bob: A key strength is the citations. Answers come with small numbered references, and clicking one jumps straight to the passage in your document that the answer is based on. That makes it useful for long reports, course reading or a dense rental contract. There's also a spoken summary option: two AI voices discuss your material in a short podcast-style conversation, which you can listen to on the commute instead of reading forty pages at your desk.

Samantha: Two important limits. First, it can still misread or get things wrong. The citations make checking easy, so use them, especially for anything involving money, health or legal deadlines. Second, your files are uploaded to Google's servers. Keep confidential work documents, medical records or anything your employer hasn't cleared out of it. The free version also comes with usage limits; heavier users can pay for higher ones through Google's subscription plans.

Bob: A good first test for this afternoon: pick one long document you've been putting off. The terms of your phone contract, a pile of school newsletters, or a work report you're allowed to share. Upload it, ask for the three most important points and any deadlines, then click the citations to see whether the summary holds up. It takes about ten minutes, and you'll know quickly whether it earns a place in your routine.

Samantha: Today we covered: AI-powered air defence switching on in Europe, South Korea's record chip exports, Meta's humanoid robot prototype built with Fourier, actively exploited flaws in GitLab and Cisco firewalls, and two things to try yourself: Microsoft PowerToys and NotebookLM.

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.