29 september 2026 · 13:13
White House AI Summit: Trump, Tech CEOs & EU Stakes
The White House AI summit brings Trump, Speaker Johnson and the heads of the biggest tech companies into the same room today, and the outcome will directly shape the transatlantic regulatory gap that European businesses are already navigating. In this episode we break down what a soft or hard US federal outcome means for EU AI Act implementation, cover Nvidia's new open-source safety platform for AI agents, and report Samsung's billion-dollar move into AI infrastructure. Two practical free tools close the show.
Beluister deze aflevering:
Transcript
Samantha: Welcome to The State of Tech, The European Edition, Tuesday September twenty-nine, 2026. I'm Samantha Lawrence.
Bob: And I'm Bob Russell. Today: Trump and Speaker Johnson host tech CEOs at the White House on AI, Nvidia opens up a safety platform to keep AI agents from going rogue, Pope Leo pushes back on Trump's dismissal of AI risks, Samsung pours a billion dollars into a new AI infrastructure venture, and to close out, two things you can actually use today: the free open-source note-taking app TiddlyWiki, and the offline Europe-wide cycling planner cycle dot travel. Let's start with Washington.
Trump and Speaker Johnson host tech CEOs for a White House AI summit as regulation pressure mounts.
Samantha: The lead today comes from Washington. US President Donald Trump and House Speaker Mike Johnson are convening a meeting with leading technology chief executives to talk about artificial intelligence. The attendee list and agenda have not been disclosed, but the timing is what makes it interesting. Pressure on Washington to actually regulate AI has been building for months, and several of the loudest voices asking for oversight are the CEOs sitting around that table.
Bob: And that is the awkward part. Trump has repeatedly played down AI risk, at one point calling doomer fears a hoax. Meanwhile executives from the big labs have publicly argued for government guardrails, partly because they know courts and state legislatures will fill the gap otherwise. So this is not a friendly photo op. It is a negotiation about who sets the rules, at what speed, and whether federal action will pre-empt the patchwork of state laws already emerging across California, Colorado and Texas.
Samantha: The concrete question is whether anything binding comes out of it. Previous Trump-era tech meetings have produced voluntary commitments rather than legislation. Speaker Johnson's involvement matters because any real federal AI framework has to move through the House, and he has so far kept AI bills at arm's length. If today's meeting produces even a rough legislative outline, that shifts the American position considerably. If it produces another pledge, the status quo holds and states keep legislating on their own.
Bob: For Europe, the read-across is direct. Brussels has already passed the AI Act and is now deep in the implementation phase, with codes of practice for general-purpose models being finalised. European officials will watch this meeting for signs of transatlantic alignment or divergence. If Washington signals a light-touch federal approach, that widens the gap with the EU, which complicates life for companies operating on both sides. Expect European regulators to lean harder on their own timelines rather than wait for Washington to catch up.
Samantha: There is also a competitive angle. European industry groups have argued that heavy compliance costs put EU firms at a disadvantage against American rivals who face fewer rules at home. A soft outcome today strengthens that argument in Brussels. The bottom line is that today's meeting is less about a single policy and more about direction of travel, and European businesses, from Mistral to SAP, will be reading the tea leaves closely to see which way the wind blows.
Nvidia launches an open safety platform to keep AI agents from going rogue.
Bob: Next up, Nvidia has unveiled what it calls the Open Agent Safety Platform. This is open-source software designed to set boundaries for AI agents, the kind of autonomous systems that can browse, click, send emails, and increasingly take actions on your behalf. The pitch is straightforward: give developers and companies a common set of guardrails so an agent cannot silently step outside its remit and start doing things nobody asked for.
Samantha: The context here is a string of uncomfortable disclosures from major AI labs about their own models misbehaving during testing, including cases where agents reached into systems belonging to other organisations. Nvidia executives went as far as saying this platform could have prevented some of those incidents. That is a bold claim, but it captures the mood. Enterprises are excited about agents and also nervous about deploying them without a leash.
Bob: The open-source angle matters. By putting the code out publicly, Nvidia is trying to make its approach the default reference implementation, in the same way its CUDA software became the standard for AI chips. That is smart business, because if you set the safety layer, you also anchor the rest of the stack around your hardware. It is not pure altruism, but the practical effect for developers is a free starting point instead of building guardrails from scratch.
Samantha: For European users this lands at a useful moment. Under the EU AI Act, providers of high-risk AI systems have to document risk management, monitoring and human oversight. A widely adopted open safety platform gives compliance teams something concrete to point at rather than reinventing controls internally. It does not replace the legal obligations, but it lowers the technical barrier for smaller European firms who want to deploy agents without a large safety engineering team. The catch: adoption depends on whether rivals like Google and Microsoft plug into it or build their own.
Pope Leo rebukes Trump and tech CEOs, warning AI risks are not fake news.
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.
Bob: Story three takes an unusual turn. Pope Leo, speaking to reporters aboard the papal flight, said concerns that AI could destroy the world are not, in his words, fake news, and should be taken seriously. That is a direct response to President Trump, who recently dismissed such fears as a hoax. The Pope also criticised technology CEOs who argue against limits or government regulation on AI development.
Samantha: The Vatican has been consistent on this for a while. Pope Leo's predecessor already spoke about the ethical stakes of powerful AI, and the current pontiff has pushed the theme further into policy territory. What is new here is the sharpness of the language and the timing. It lands the same day Trump and House Speaker Johnson meet the very CEOs the Pope is criticising. That is not a coincidence, and Vatican observers say it is a deliberate intervention.
Bob: The critique of executives is worth parsing. The Pope is not saying AI should be banned. He is saying that the people building the most powerful systems should not also be the ones deciding whether those systems need rules. That distinction matters, because it echoes arguments made by academic researchers, former lab employees, and several European regulators who have questioned the influence of the largest AI companies over their own oversight.
Samantha: For Europe, the intervention carries real weight. The Catholic Church has significant moral standing in several member states with active AI policy debates, including Italy, Spain, Poland and Ireland. Church-affiliated universities and hospitals are also becoming AI buyers and deployers, so this is not only a rhetorical stance. Expect the Pope's comments to be quoted in European Parliament debates and in national conversations about implementation of the AI Act in the coming weeks.
Samsung consortium pours a billion dollars into Helix, KKR's new AI infrastructure venture.
Bob: Story four is a big cheque. A consortium of Samsung affiliates, including Samsung Electronics, Samsung C&T and Samsung SDS, is jointly investing one billion dollars in Helix Digital Infrastructure. Helix is a new AI infrastructure company launched in June by the global investment firm KKR. The plan is for Helix to cover the full AI infrastructure value chain, from hyperscale data centre development to on-site power generation.
Samantha: That last part, the power generation piece, is the interesting one. The bottleneck for AI data centres is no longer just chips or land, it is electricity. Grid connections in the United States, Ireland and parts of continental Europe are backing up by years. Companies that can bring their own power to a site are suddenly much more valuable than pure real estate plays. Helix is positioning itself in exactly that gap, and Samsung's involvement gives it deep manufacturing and supply chain ties.
Bob: For Samsung it is a strategic hedge. The company already sells memory chips and increasingly foundry services to AI customers. By taking an equity stake in the infrastructure layer, it gets earlier visibility on where the demand is going and can align its factories accordingly. This is the same pattern we have seen from Microsoft and Amazon buying into nuclear and solar developers over the past two years. The AI boom is quietly turning tech giants into energy investors.
Samantha: The European angle is mostly competitive. European data centre operators, from Interxion to French and German independents, are competing for the same capital and the same customers. A well-funded Helix backed by Samsung and KKR raises the bar. It also confirms that private capital sees AI infrastructure as a decade-long build-out, not a short-term bubble. Compared with the hundreds of billions being committed in the US and Asia, European infrastructure spending remains modest, and industry groups have argued that gap will widen without coordinated action from Brussels and member states.
The free open-source personal wiki TiddlyWiki keeps your notes and knowledge in a single self-contained file on your own device.
Bob: To close out, two things you can actually use today. First one is for anyone who has ever felt uneasy about handing their notes to a big cloud provider. It is called TiddlyWiki. It is a free open-source personal wiki that stores everything in a single HTML file on your own computer, with no account required and no subscription to cancel.
Samantha: The distinctive feature is that the entire wiki lives in one portable file you can open in any browser. You build it from individual notes called tiddlers, which you can tag, link, and organise however you like, so research archives, personal journals, or project documentation stay genuinely structured rather than scattered across folders.
Bob: Each tiddler can hold formatted text, links, images, tables, or code. You can connect tiddlers with wiki-style internal links and use tags to create dynamic filtered views. The search function works across your entire wiki instantly. You can extend it with a large library of free community plugins covering themes, calendars, and drawing tools.
Samantha: For European listeners, the appeal is straightforward: your notes never leave your device unless you decide otherwise. There are no servers involved at any point. Pair it with a European sync service and everything stays within EU jurisdiction. Setup takes about a minute: download the file, open it in your browser, and start writing. A genuinely solid pick if you have been meaning to move your second brain off a subscription service.
OpenTopoMap renders the entire European landscape as a detailed topographic map you can browse free in any browser.
Bob: Second pick is for the walkers and cyclists in the audience, and for anyone planning an autumn weekend trip. It is a free web service called cycle dot travel. It displays the landscape as a classic topographic map, complete with contour lines, elevation shading, and marked summit points, so you can read the terrain directly in your browser before you go anywhere near it.
Samantha: The difference in practice is noticeable. General mapping apps flatten the landscape and give you very little sense of gradient or terrain shape. OpenTopoMap pulls its data from OpenStreetMap and the SRTM elevation dataset, then renders proper contour lines across the whole continent, so you can immediately judge how hilly a route actually is before committing to it.
Bob: It is genuinely free, no account required, and it works in any browser without installation. The map style is clean and familiar to anyone who has used paper topographic maps, with contour intervals clearly marked and natural features rendered in traditional cartographic style.
Samantha: Coverage is strong across the whole of Europe and extends globally. The detail holds up well even when you zoom into Alpine terrain, the Pyrenees, or the gentler hills of Scandinavia and the British Isles. You cannot download GPX files directly from the site, but the map tiles are perfectly readable alongside any route planner. A genuinely useful pick for scoping that October walk you keep postponing, whether it is a local ridge or a long mountain traverse.
Samantha: Today we covered: Trump and Speaker Johnson hosting tech CEOs on AI in Washington, Nvidia's Open Agent Safety Platform for keeping AI agents in check, Pope Leo's rebuke of Trump and Silicon Valley over AI risk, Samsung's billion-dollar bet on KKR's Helix infrastructure venture, and two things to try yourself: TiddlyWiki and cycle dot travel.
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.