Google makes AI mandatory in search, EU backs chip design

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

Bob: And I'm Bob Russell. Today: Google makes AI a fixed part of search, voters push back against the AI building boom, Europe puts five hundred and twenty-five million euros into chip startups, Meta buys the AI coding startup Conductor, and two things to try yourself: the private AI chat service Duck.ai and the food scanner app Open Food Facts. Let's start with Google.

Google makes artificial intelligence a mandatory part of search for everyone who uses it.

Samantha: Google has announced that artificial intelligence is now a mandatory part of its search experience. It is not an extra tab or an experiment you opt into. According to the announcement, AI is being woven into the core of how search works, so the way people ask questions and get answers on the world's most used search engine changes at the root. Google says the goal is more sophisticated and more personalised results. For a lot of people, search is the front door to the internet, so this touches almost everyone.

Bob: The word that stands out is mandatory. Until now, AI in search has mostly been presented as a layer on top of the familiar list of links. This announcement frames it as a basic building block instead. The source does not say whether users will keep any way to switch the AI parts off, or what the classic results will look like next to them. Those details will decide how big the change feels for someone typing in a train time or a recipe.

Samantha: There is also a strategic signal. Google is putting its most advanced AI directly into its most widely used product, rather than keeping it in separate apps. That shows where the company sees its future: not AI as a side product, but AI as the product itself. It also raises the stakes for rivals. If search by default means an AI-shaped answer, competitors can either offer something comparable, or position themselves as the opposite. Both directions are now a clearer market position than they were a week ago.

Bob: For European users the questions are concrete. Personalised AI answers depend on what the system knows about you, so data privacy comes straight into view. And when an AI summarises the web for you, it becomes harder to see why you get this answer and not another one. That is the question of algorithmic transparency. The EU has its own rules on data protection and on large online platforms, and how this rollout fits within those frameworks is something the announcement itself does not address.

Samantha: The bottom line is that the search box many Europeans use dozens of times a day is changing character. For businesses that depend on being found through Google, from small web shops to news sites, the way information gets surfaced matters a great deal, and the announcement gives no detail yet on how links and sources will be shown. Key takeaway: watch how this lands in Europe in the coming weeks, both on screen and with regulators. Later in this episode, one alternative for people who want AI on their own terms.

Voters approve data centre moratoriums and elect candidates who want to slow the AI buildout.

Bob: Voters have sent the AI industry a message at the ballot box. In elections on November four, according to the report, voters in several places approved moratoriums on new data centres, effectively pausing their construction, and elected candidates who want to slow down the AI building spree. That spree is estimated at around ten trillion dollars worth of infrastructure. The report calls it an AI earthquake. The core point: many voters see the race for ever more AI computing power not as progress, but as a threat to local economies and communities.

Samantha: The money and the mood are pulling in opposite directions. Tech companies are planning enormous investments in data centres, and these results suggest local communities are increasingly willing to say no. The report describes the buildout as resource-intensive, and that is exactly where the AI boom meets everyday life: in a specific town, on a specific plot of land. For tech leaders and investors, the report frames this as a clear warning. Political risk is now part of the AI infrastructure story.

Bob: On the same day, pressure came from another direction too. US Representative Darrell Issa warned Congress that it needs to act on AI immediately. He pointed to fragmented federal oversight as an unacceptable risk to national security and economic competitiveness, with AI moving fast into defence, healthcare and finance. He also warned that delays could leave the US vulnerable as other nations build comprehensive AI rules. So voters want to slow the buildout locally, while a lawmaker warns about falling behind. Two very different worries, landing on the same day.

Samantha: For Europe, the report's analysis suggests this backlash could inspire similar movements here. That could mean more scrutiny and more regulatory hurdles for AI infrastructure projects in European regions. Issa's call could also feed into global policy discussions, possibly encouraging closer cooperation between the US and Europe on AI governance standards. Key takeaway: the question of where AI gets built, and who gets a say, is no longer only a boardroom decision. It is now showing up at the ballot box.

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.

The European Union puts five hundred and twenty-five million euros into homegrown chip design startups.

Bob: Europe is putting five hundred and twenty-five million euros into chip startups. The European Union announced the plan on November four. Four hundred and forty million goes to grants, and eighty-five million to a design platform. The aim is to help young companies grow faster and make Europe less reliant on external suppliers. And this one is not about chatbots. It is about semiconductors, the chips that everything digital runs on.

Samantha: The focus is chip design, and one number explains why. European companies account for less than one percent of worldwide revenue in chip design. Design is the stage where the blueprint of a chip is drawn up, before anything gets manufactured. So Europe is starting from a very small base in exactly the part of the chain this money targets.

Bob: EU tech chief Henna Virkkunen tied it to the bigger picture. Her argument is that a vibrant design ecosystem is crucial for boosting domestic manufacturing and the entire European value chain. Put simply: if more chips are designed in Europe, there is more reason to build them in Europe too. The eighty-five million euro design platform follows that same logic, supporting the design stage itself.

Samantha: For European startups, grants mean money that does not have to come from investors, which changes how fast a young company can move. For listeners, the effect is slower and less visible than a new app, but it shapes where the technology inside everyday devices comes from, and how dependent Europe stays on suppliers elsewhere.

Bob: Some perspective on scale. Half a billion euros is a substantial sum for a group of startups, but the starting point is less than one percent of global design revenue. Notable is that the EU is targeting the design stage specifically, rather than only factories. The bottom line: Brussels is betting that growth in design will pull the rest of Europe's chip industry along with it.

Meta buys AI coding startup Conductor and shuts its service down within weeks.

Samantha: Meta has bought Conductor, a young AI coding startup founded in 2024. Conductor lets software developers put several AI coding helpers to work at the same time, each in its own online workspace, so they can build different parts of a program in parallel. The acquisition was confirmed on November four, and the technology moves into Meta's Superintelligence Labs. The goal is stronger capabilities in what is called agentic AI: AI that does not just answer questions, but carries out tasks on its own.

Bob: For current users, the most concrete detail is a deadline. Conductor's cloud service shuts down on November twenty, about two weeks from now. Developers who rely on it will have to find an alternative. Founder Charlie Holtz is joining Meta, and he is already teasing a new product interface by early 2027, meant to integrate AI more deeply. So the product people know disappears, while something new is promised for next year, without details yet on what it looks like or who it is for.

Samantha: What Meta is buying here is less a finished product and more a way of working: many AI agents running side by side on the same software project. Bringing that into its Superintelligence Labs shows Meta is aiming for agents that do real work, not just chat. And the teased interface for early 2027 suggests Meta wants its own entry point for developers. For a company of Meta's size, adding a small specialised team is a fast way to get there.

Bob: For European developers it cuts both ways. On one side, this could eventually mean more advanced AI coding tools from Meta for European programmers. On the other, it is another case of AI talent and technology being consolidated under a large US tech company. And as Conductor shows, when a small tool is absorbed, users can lose the product they chose, here with roughly two weeks' notice. Key takeaway: in AI tooling, today's startup can be tomorrow's acquisition, and the service can disappear with it.

Duck.ai lets you chat with several AI models without an account, for free.

Samantha: To close out, two things you can actually use today. First, and it fits with our Google story: Duck.ai, the AI chat service from the privacy-focused search company DuckDuckGo. You open it in any browser, or inside DuckDuckGo's own apps, and you can pick from several well-known AI models made by different companies, all in one place. No account is needed to start. The free version comes with usage limits, and there is a paid subscription for heavier use and more advanced models.

Bob: The selling point is privacy. According to DuckDuckGo, it removes your IP address, the number that identifies your internet connection, before passing your question on to the AI provider. Recent chats are stored on your own device rather than on DuckDuckGo's servers. For people who want to use AI without tying every question to an account, that is the appeal.

Samantha: A practical way to try it this afternoon: take a question you would normally type into a search engine, like planning a weekend trip or comparing two phone contracts, and ask it in Duck.ai. Then start a new chat with a different model and ask the same thing. You will quickly see whether the answers differ in tone, detail and accuracy, a fast lesson in how much the model you pick shapes what you get. It takes about five minutes, it costs nothing, and it works on laptop and phone.

Bob: Now the limitations. The privacy protection rests on DuckDuckGo's own promises and its agreements with the AI companies, so it is worth reading the privacy policy before typing anything sensitive. Like every AI chat, it can confidently get things wrong, so check important facts elsewhere. The list of available models also changes over time, so what you see may differ from older reviews. The bottom line: a free, low-effort way to see what different AI assistants can do, without signing up for each one.

Open Food Facts scans grocery barcodes and shows what is really inside your food.

Samantha: The second one has nothing to do with AI: Open Food Facts, a free app and website that shows what is in your groceries. You scan the barcode on a product with your phone camera and get the ingredients, allergens and additives, plus the Nutri-Score, the colour-coded A to E nutrition label, and a rating for how heavily processed the food is. It is run as a non-profit project that started in France, and the app is free on Android and iPhone.

Bob: What makes it different is how the data gets collected. The database is built by volunteers and users, a bit like Wikipedia for food products. If you scan something that is missing, the app asks you to photograph the packaging, the ingredients list and the nutrition table, so you can add it yourself. The data is openly licensed, which means other apps and researchers can build on it too.

Samantha: A concrete thing to try today: open the app, scan five products from your kitchen cupboard, say a breakfast cereal, a pasta sauce and a snack, and compare their scores. It is a quick way to see whether the packaging and the numbers tell the same story. It takes about ten minutes. For households dealing with allergies, the ingredient overview can also be a handy first check while shopping.

Bob: The limitations matter here as well. Because the data comes from volunteers, entries can be incomplete or out of date, for example when a manufacturer changes a recipe. So for allergies, the label on the package itself remains the thing to check. And the scores are a summary, not dietary advice. Bottom line: Open Food Facts puts a lot of food information in your pocket for free, and every product someone adds makes it more useful for the next shopper.

Samantha: Today we covered: Google making AI mandatory in search, voters pushing back against the AI building boom, Europe's half-billion-euro bet on chip design, Meta buying Conductor, and two things to try yourself: Duck.ai and Open Food Facts.

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.