Bank of England AI Systemic Risk Warning: July 2026

Samantha: Welcome to The State of Tech, The European Edition, Saturday July eleven, 2026. I'm Samantha Lawrence.

Bob: And I'm Bob Russell. Today: the Bank of England sounds the alarm on AI risks in finance, the White House pours over two billion dollars into quantum computing, a Chinese robotics maker raises three quarters of a billion for wheeled humanoids, a fresh wave of AI PCs and workstations hits the market, and to close out, two things you can actually try today: the free open-source password manager KeePassXC, and the AI writing assistant NovelCrafter. Let's start with the Bank of England.

The Bank of England warns that the AI boom is quietly reshaping the risk map of global finance.

Samantha: The Bank of England has just fired one of the loudest warning shots yet about artificial intelligence in the financial system. In its latest half-yearly stability review, the central bank says the rapid rollout of AI across banks, insurers, and trading floors is opening up new avenues for cyberattacks and could amplify shocks across the market. The concern is not one dramatic incident. It is the sheer surface area: more AI systems, more data flows, more third-party providers, all interconnected.

Bob: And that last point is where the Bank is putting real pressure. A handful of critical technology providers now sit underneath a huge share of Europe's financial plumbing. If one of them gets compromised, or an AI model starts behaving in an unexpected way under stress, the effects can ripple fast. The regulator is pushing firms away from the old checklist approach to security and towards continuous cyber resilience, meaning constant testing, monitoring, and assuming that something, somewhere, will go wrong.

Samantha: There is also a market-behaviour angle that deserves attention. When many banks and hedge funds start using similar AI models to make trading and risk decisions, their reactions to a shock can start to look almost identical. That kind of herd behaviour, driven by machines rather than people, can turn a mild wobble into a much larger sell-off. The Bank of England is essentially asking whether regulators fully understand how these models will behave together when markets get rough.

Bob: For European institutions this lands right on top of existing rules like DORA, the bloc's digital operational resilience framework for finance. Expect closer scrutiny of AI vendors, tougher stress tests that include AI-specific scenarios, and more questions from supervisors about who exactly is running the models behind key decisions. The bottom line is that AI in finance is no longer just an efficiency story. Central bankers are starting to treat it as a systemic issue, and that changes the compliance bill for every bank across the continent.

The White House pledges over two billion dollars to push American quantum computing to utility scale.

Samantha: Washington has put serious money behind quantum computing. The White House announced more than two billion dollars in incentives for nine American quantum-focused companies, tied to a broader push to build what officials call utility-scale, fault-tolerant quantum computers. The package is linked to a benchmarking initiative run by the US defence research agency DARPA, which is trying to cut through the hype and figure out which approaches can realistically deliver useful machines by 2033.

Bob: The framing here matters. For years, quantum has lived somewhere between physics lab and investor pitch deck. What the US is now trying to do is force a data-driven reality check: which qubit designs actually scale, which error-correction schemes hold up, and which companies are worth backing with public money. That is a shift from funding science to picking industrial champions, and it puts pressure on every other region that wants a seat at the table.

Samantha: Europe already has serious quantum research, from Delft to Munich to Grenoble, and startups like Pasqal, IQM and eleQtron are building real hardware. But funding here is fragmented across national programmes and the EU Quantum Flagship, and it does not come close to two billion in one go. Industry groups have been arguing for months that Europe needs a much bigger, coordinated push if it wants to stay in the race for the machines that could eventually crack today's encryption and accelerate drug discovery.

Bob: There is also a security dimension European policymakers cannot ignore. A future fault-tolerant quantum computer, wherever it is built, could in principle break widely used encryption. That is why regulators are already nudging banks, telecom operators and government agencies towards post-quantum cryptography. Today's announcement is a reminder that the clock on that migration is not theoretical. Somebody is spending real money to make it happen.

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.

Chinese robotics maker AI² raises seven hundred thirty-five million dollars for wheeled humanoid robots.

Bob: Shenzhen-based AI² Robotics has closed a funding round of around seven hundred thirty-five million dollars, pushing its valuation past three billion. The backers are a striking mix of state-linked entities, industrial companies, and financial investors, which tells you how strategic China now considers this field. AI² is not building the walking, running humanoid you see in flashy demo videos. Its robots have a humanoid torso and five-fingered hands, but sit on wheels rather than legs.

Samantha: That design choice is more important than it sounds. Legs are hard, expensive, and often unnecessary in warehouses, factories, or hospitals where the floor is flat. By going with wheels, AI² is aiming at machines that are simpler to build, cheaper to maintain, and easier to deploy at scale. The company frames it as prioritising mechanical simplicity and durability, which in practice means faster payback for the businesses buying them.

Bob: And that is where the competitive pressure on Europe becomes real. Chinese robotics firms are now getting the kind of capital that lets them iterate hardware quickly and undercut on price. European players in industrial robotics, from ABB to KUKA to newer startups, have deep expertise but often work with tighter funding rounds. If wheeled humanoids from China start arriving in European logistics centres at a fraction of the cost of Western equivalents, procurement decisions will follow the price tag.

Samantha: There is a policy angle brewing too. Brussels has already been debating how to handle Chinese electric vehicles, solar panels, and now potentially industrial AI hardware. Robotics of this kind touches sensitive areas: factory data, workplace surveillance, and critical infrastructure. Expect questions in the coming months about certification, data flows, and whether European buyers should be nudged towards domestic alternatives, even if that means paying more.

ASUS and Dell push a fresh wave of AI PCs and workstations built around on-device neural chips.

Bob: The AI PC story took another step this week. ASUS launched new Vivobook laptops in India built around Intel processors with a dedicated neural processing unit, essentially a small chip inside the laptop that handles AI tasks locally. Dell followed with a new line of Pro Precision workstations aimed at engineers, designers and researchers, along with a deskside AI machine pitched at teams that want to run larger models without touching a cloud provider.

Samantha: The interesting shift is not any single product. It is that the industry is quietly rewiring what a normal computer looks like. For years, the CPU and graphics chip did all the heavy lifting. Now there is a third component whose only job is to run AI models efficiently, on your desk, without sending data anywhere. That means faster transcription, smarter search across your own files, and image and video tools that do not need a subscription to a cloud service.

Bob: For European buyers this matters on two fronts. On price, these machines are starting to appear in the mid-range, not just at the very top, which will accelerate refresh cycles in offices next year. On data protection, running AI locally is genuinely attractive under the GDPR: sensitive documents, HR files, or client data can be processed on the device rather than shipped off to a US or Chinese cloud. That is a real selling point for law firms, hospitals, and public sector teams.

Samantha: The catch is software. The hardware is arriving faster than the applications that fully use it. Most everyday programs still lean on the cloud for their AI features, which means many buyers will pay for a neural chip they barely touch in year one. Over the next twelve months, whether Microsoft, Adobe and the big European software vendors actually rebuild their tools to run locally will decide whether the AI PC is a genuine leap, or an expensive sticker on the lid.

A free password manager that finally kills the sticky note under the keyboard.

Bob: To wrap up, two things you can actually try this weekend. The first is KeePassXC, a free open-source password manager that has been quietly winning fans among people who do not want to trust a big cloud provider with their entire digital life. It runs on Windows, macOS and Linux, and there are companion apps that work smoothly on iPhone and Android through compatible clients.

Samantha: The idea is straightforward. KeePassXC stores all your passwords in a single encrypted file on your own device. You unlock it with one strong master password, and from then on it fills in logins, generates long random passwords for new sites, and warns you when you are reusing the same one across services. Because the file lives on your machine, nothing is sent to a company's servers unless you actively choose to sync it yourself, for example through your own cloud drive.

Bob: For European users worried about data location, that self-hosting model is a big part of the appeal. There is no subscription, no account to create, and no vendor that could be breached and leak your vault. Security researchers regularly praise the project for being transparent and well maintained, and it has been audited multiple times over the years. Setup takes about fifteen minutes: install the app, create a database, set a strong master password, and start importing logins from your browser.

Samantha: One practical tip from long-time users: back up that encrypted database file somewhere safe, like an external drive or your own encrypted cloud folder. If you lose it, there is no support desk that can recover your passwords, which is exactly the trade-off that makes it private in the first place. If you have been meaning to escape the browser's built-in password saver, this is a weekend project that pays off for years.

An AI writing assistant built specifically for people working on long, structured projects.

Bob: The second find is NovelCrafter, an AI-powered writing tool that has been getting a lot of attention from authors, screenwriters, and anyone tackling a long-form project. Unlike a general chatbot, it is built around the way a book or screenplay is actually structured: chapters, scenes, character sheets, world-building notes, and a timeline that keeps everything in sync.

Samantha: The clever part is how it uses AI. You write your own text, and the assistant works in the margins: suggesting continuations when you are stuck, keeping track of who knows what at any point in the story, and flagging when a character's eye colour suddenly changes in chapter twelve. Users on writing forums often describe it as a research assistant and continuity editor rolled into one, rather than a machine that writes the book for you.

Bob: It runs entirely in the browser, so there is nothing to install, and it works on any laptop. There is a free tier that lets you try the core features, and paid plans that unlock more AI credits and let you plug in your own preferred model, including options from Anthropic and OpenAI. For European users, the ability to choose which model handles your text is useful if you care about where your work is processed.

Samantha: A quick word of realism: reviewers consistently point out that the AI helps most when you already have a clear idea of what you are writing. It is not a magic bestseller machine. But for anyone staring at a half-finished manuscript, a thesis, or a company report that never seems to end, it is worth an hour on a Saturday afternoon to see whether the structured approach clicks. And that is your two to try this weekend.

Samantha: Today we covered: the Bank of England's warning on AI risks in finance, Washington's two billion dollar bet on quantum computing, AI² Robotics raising three quarters of a billion for wheeled humanoids in China, the new wave of AI PCs and workstations from ASUS and Dell, and two things to try yourself: KeePassXC and NovelCrafter.

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.