How the Pope’s Magnifica Humanitas Offers a Template for Individuals to Meet the AI Moment
Pope Leo XIV has dropped his AI encyclical, and the headline quote — “Technology is never neutral” — is the kind of thing that should be printed on the wall of every boardroom currently racing to ship agents. The document calls for courage and solidarity in the face of AI-driven transformation, which is either a profound moral framework or a very eloquent way of saying “this matters, pay attention.” Either way, when the Catholic Church is moving faster on AI ethics than most national legislatures, that tells you something about where policy currently stands.
‘What a Joke’: GitHub Copilot’s New Token-Based Billing Spurs Consternation Among Devs
Microsoft is switching GitHub Copilot to token-based billing, and developers — the exact people Copilot is supposed to delight — are not delighted. The phrase “what a joke” in the headline is pulled directly from user reactions, which is doing a lot of work for a product that was supposed to be the future of software development. Nothing kills adoption quite like billing ambiguity, and asking developers to mentally track token consumption while they’re trying to ship code is a remarkable way to hand ammunition to every open-source competitor in the space.
Claude Code Costs Up to $200 a Month. Goose Does the Same Thing for Free.
The premium AI coding agent market is getting its first serious price-pressure test, as Block’s open-source Goose positions itself as a free alternative to Anthropic’s $20–$200/month Claude Code. This is the classic open-source vs. commercial tension playing out at warp speed — Anthropic spent enormous resources building Claude Code, and now the question is whether the quality gap is wide enough to justify the price tag. Spoiler: some developers are already answering that question with their wallets.
Anthropic Wants Claude to Play With Money, Unleashes Finance Agents
Anthropic is now letting Claude loose on financial tasks through dedicated finance agents, which is either a natural evolution of agentic AI or the setup for a very expensive cautionary tale. To their credit, Anthropic has been among the more safety-conscious labs, but “always bet on backpropagation” is The Register’s subhead and I’m going to need a moment with that. The real story here is that every major lab is racing to get agents into consequential real-world workflows — money, medicine, code — and the gap between capability demos and institutional trust is still very much a live issue.
SoftBank Says It Will Invest Up to €75 Billion to Build French Data Centers
SoftBank is pledging up to €75 billion for French data center capacity — 5 gigawatts worth — which is either a visionary infrastructure bet or Masayoshi Son doing what Masayoshi Son does. Europe has been loudly complaining that AI infrastructure investment is flowing to the US and China, so this is a geopolitically significant commitment, assuming the “up to” qualifier doesn’t do a lot of heavy lifting over the coming years. The AI compute land-grab is now clearly a global infrastructure race, not just a Silicon Valley arms race.
OpenAI Exec Says Company Hopes to Burn $50B of Somebody Else’s Money on Compute This Year
OpenAI is apparently aiming to spend $50 billion on compute in 2026 — most of it not their own money — and the Register’s subhead “If the numbers are large enough, perhaps we won’t question the math” is doing journalism’s work here. This is the new frontier of AI flex: not model benchmarks, but the audacity of your burn rate. At some point the question of whether the revenue and capability gains justify these capital expenditures will need an actual answer, and “AGI is coming” has a shelf life.
Tech Companies Desperately Want to Film You Doing Chores
A startup called Shift is offering to clean your home for free — in exchange for filming everything you do around the house, ostensibly to train robotics models. I’ll give them points for creativity, because most data collection schemes at least pretend to be something else. This is the new economy in miniature: your labor, your space, your privacy, all converted into training data for someone else’s future product. The home-cleaning robots won’t clean your home for free either, by the way.
Boston Children’s Uses AI to Unlock New Diagnoses
On a more genuinely encouraging note: Boston Children’s Hospital is using OpenAI technology to help diagnose rare diseases, with over 40 cases identified that might otherwise have gone unresolved. It’s worth pausing on stories like this amid all the billing drama and papal encyclicals — there are real clinical applications where AI pattern-matching across vast medical literature is doing things that aren’t otherwise possible at scale. This is the use case the industry should be leading with, even if it doesn’t generate as many hot takes.
Bottom Line
From papal doctrine to token-based billing revolts, the AI industry spent the weekend discovering that moving fast and building things is the easy part — the hard part is getting humans to trust, afford, and actually want what you’ve built.