1. Your job is making you anxious. ChatGPT isn't helping
If 21st-century capitalism has one central tenet, it's that innovation rules and disruption drives progress. And generative AI is one of the most disruptive innovations of our lifetime. Which is why, as Morra Aarons-Mele writes in HBR, it's making us all so bloody anxious.
The anxiety is entirely rational. Our nervous system is designed to react to sudden change and perceived threats, and AI is delivering stressors at both scale and speed. Aarons-Mele identifies three drivers: lack of control over the speed of change, loss of meaning in work, and uncomfortable emotions we'd rather avoid. That last one is particularly acute when your boss casually refers to ChatGPT as the "Chief Marketing Officer" in front of the actual CMO.
The article's strength is in treating AI anxiety not as something to be suppressed, but as information to be understood. What values are being threatened? Is it fairness? Trust? Craft? Agency? Understanding your emotional response, Aarons-Mele argues, is the key to acting strategically rather than reactively. As Brené Brown puts it: "If you're not feeling unsettled, you're probably not paying attention."
2. Elon Musk reinvents Wikipedia, makes it worse
Elon Musk's latest wheeze is Grokipedia, an AI-powered "encyclopedia" that's supposed to be a superior alternative to Wikipedia. Spoiler: it isn't. Instead of addressing Wikipedia's perceived biases, it simply reinforces Musk's own worldview, whilst lacking the reliability that makes Wikipedia actually useful. Read more at the FT.
The fundamental problem, as Jemima Kelly points out, is that Grokipedia illustrates the ongoing challenges of maintaining truth and accuracy in AI-generated knowledge without proper human oversight. Wikipedia works not despite its human editors, but because of them. It's a system built on transparency, sourcing, and endless arguments about whether something is notable enough to include.
What Musk has created is the opposite: an opaque black box that produces plausible-sounding text with all the hallmarks of AI-generated slop. It's telling that someone with Musk's resources and apparent obsession with "free speech" can't grasp that knowledge curation requires exactly the kind of human judgment and community governance he seems to despise. Another win for the "move fast and break things" crowd.
3. Microsoft and OpenAI's divorce just got very expensive
OpenAI is ending October with a new for-profit structure, a new deal with Microsoft, and an entirely new level of pressure to achieve artificial general intelligence. The Verge has the details.
The stakes are enormous. The 2019 partnership between OpenAI and Microsoft included an "AGI clause" that said Microsoft's rights to use OpenAI's technology would end once OpenAI achieved AGI. But now everything has changed. Microsoft can independently pursue AGI alone or with third parties, and crucially, can use OpenAI's IP to do it. An independent panel will now verify any AGI declaration, rather than OpenAI making unilateral decisions.
This restructuring puts billions of dollars on the line and kicks off an arms race in earnest. Microsoft could work with Anthropic or other OpenAI competitors to reach AGI first. Meanwhile, Sam Altman keeps shifting the goalposts, recently saying AGI is "hugely overloaded" as a term. What's clear is that the race to define and achieve AGI is now as much about corporate competition as it is about technological capability.
4. Germany's escape from Outlook is going exactly as well as you'd expect
The German state of Schleswig-Holstein has completed its migration from Microsoft Exchange and Outlook to open source alternatives, moving over 40,000 mailboxes with more than 100 million emails to Open-Xchange and Thunderbird. Heise reports on what Minister Dirk Schrödter calls a milestone for digital sovereignty.
This isn't just about email. The state is systematically replacing its entire Microsoft stack: LibreOffice is replacing Office, Nextcloud is replacing SharePoint, and eventually Linux will replace Windows across all state computers. It's a genuinely ambitious project, and one that hasn't been without problems — Schrödter recently had to admit to errors and downtime during the migration.
But here's the thing: they're doing it anyway. The northern German state is betting that independence from large tech companies is worth the pain of transition. And unlike most digital sovereignty projects, this one is actually happening, with real deadlines and real consequences. Other European governments are watching closely.
5. Cory Doctorow says Europe has built its own Berlin Wall (and he's right)
Cory Doctorow has written a brilliant piece arguing that Europe's digital sovereignty efforts will fail without repealing Article 6 of the EU Copyright Directive. His point? You can build all the European alternatives you want, but if it's illegal to create the tools to migrate away from US platforms, you've achieved nothing.
The anti-circumvention laws the EU adopted under US pressure create a legal barrier — a "Berlin Wall," as Cory calls it — that prevents European companies from building migration tools. The Digital Markets Act requires interoperability, but it relies on gatekeepers' goodwill. We've seen how that works with Apple's malicious compliance.
Cory's insight is that this isn't just about consumer rights or repair anymore. It's about whether Europe can ever break free from US tech dependency. Building a European cloud stack is pointless if your citizens and businesses can't legally extract their data from Microsoft, Google, and AWS. The EuroStack Initiative is taking note.
6. Francis Fukuyama would like to pop Silicon Valley's AGI balloon
Francis Fukuyama has written a thoughtful piece puncturing the AI hype bubble. His argument is simple: even if we achieve superintelligent AI, it won't deliver the explosive economic growth that Silicon Valley promises because intelligence isn't the binding constraint on growth.
The real constraints are material and political. We're already running into planetary limits. China, America, and Europe, at 10 per cent annual growth, would rapidly exhaust agricultural land, water, energy, and everything else. At the micro level, translating smart ideas into physical products requires iterative testing in the real world, which no amount of intelligence can simulate.
But the killer point is about implementation. AGI might know how to provide clean water to a struggling city in the developing world. But the problem isn't knowledge — it's the political realities of vested interests, corruption, and armed water mafias. Intelligence doesn't overcome those obstacles. It's a useful corrective to the cathedral of genius worship that Silicon Valley has become.
7. I let AI buy me a bookcase, and now I have questions
Claer Barrett at the FT has been experimenting with AI shopping assistants, and unlike most AI hype, this actually sounds useful. She describes using an AI app to find a white bookcase, specifying exact requirements, including that she hates DIY and wants it fully assembled.
Within seconds, she had a comparison table of UK retailers, prices, delivery times, and even a summary of customer service reviews. Adobe reckons more than half of US shoppers will be using some form of GenAI by the end of this year. The next phase is "agentic commerce" — AI not just recommending products but actually completing transactions.
Shopify has partnered with OpenAI to enable merchants to sell directly through ChatGPT conversations. As these apps collect data about our lives, they'll start anticipating our needs and suggesting purchases. Retailers are terrified about what happens to brand loyalty when AI intermediates every purchase. But if an AI shopping agent could check my bank balance and arrange delivery on a day I'm working from home? I might be persuaded.
8. Canva gives away Affinity. There's a catch, obviously
Canva has scrapped the three separate Affinity apps and combined them into a single free Mac app. On the surface, this is good news — no subscriptions for core functionality, and the apps aren't being asset-stripped or rebranded.
But there are caveats. The three apps (Publisher, Photo, and Designer) are now just modes within a single app, accessed via Vector, Pixel, and Layout buttons. Users will need to heavily customise toolbars to recreate their workflows. The iPad version won't arrive until 2026. Existing Affinity documents need to be updated to open in the new app, so they won't work in the old software anymore.
Still, compared to Adobe's relentless subscription price hikes, a free one-off download is a relief. Canva clearly sees this as its route to the professional market, and presumably expects to monetise through its paid AI tiers. For users who loved the original Affinity trio, it's bittersweet — the apps aren't dead, but they're not quite the same either.
9. Microsoft has run out of plugs for its AI chips
Microsoft CEO Satya Nadella has revealed something that should worry us all: the company has GPUs sitting in inventory that it can't plug in because there isn't enough electricity. The problem isn't a shortage of chips, but a shortage of power to run them.
This isn't some theoretical future concern. AI data centres are already causing consumer energy bills to skyrocket in the US. OpenAI is calling on the federal government to build 100 gigawatts of power generation capacity annually, framing it as a strategic asset in the race against China. Meanwhile, Beijing is apparently miles ahead in electricity supply thanks to massive investments in hydropower and nuclear power.
This demand for power cannot be endless. We're already butting up against planetary limits, and if we're planning to meet AI's insatiable appetite for electricity by burning more fossil fuels, we're trading fast computers for an unliveable climate. The AI industry is effectively asking us to choose between their profit margins and a habitable planet. And right now, it looks like they're winning that argument.
10. Tech bros have been accidentally poisoning themselves for years
Silicon Valley's joyless digital monks, subsisting on meal replacement drinks to maximise productivity, have been accidentally dosing themselves with toxic lead and cadmium. Consumer Reports found that Huel's Black Edition powder contains 6.3 micrograms of lead per serving — 13 times the daily recommended limit — plus double the safe daily serving of cadmium.
Long-term lead exposure causes kidney dysfunction, hypertension, nervous system damage, and decreased cognitive performance. Cadmium causes cumulative nervous system damage. If you've ever wondered why some tech bros have such bizarre views and erratic behaviour, chronic heavy metal poisoning that impairs cognitive function might be part of the explanation.
They've been optimising themselves into toxicity, reducing the "messy realities of existing in a body" to industrial slurry. The same naive confidence that code can solve everything has led them to replace food with brain toxins. It's the perfect metaphor for Silicon Valley hubris.