Microsoft is offering buyouts to around 60,000 staff two days before it has to show investors the AI spend is paying off.
Meanwhile Apollo's numbers say the other 493 companies in the S&P 500 have seen basically nothing from AI yet - flat revenue per worker, flat margins.
I think the first AI "productivity gain" most firms report will be headcount, not output. Not quite the same thing.
£120m more for the AI Security Institute. £60m for a National Cyber Shield. £70m for a vulnerability centre. All announced today, two days before the Budget.
The Dutch government wants around 25,000 fewer civil servants by 2035, roughly 16% of central government, with AI picking up a chunk of the admin.
BUT it's not just AI - they also want to scrap or simplify at least 500 rules a year and cut advisory bodies. I think that's the step most AI plans skip. Automating a messy process just gets you a faster mess...
Simplify, then automate.
news.bloomberglaw.com/artifi…
Grok Super doesn't really seem like much of an upgrade from Grok Premium? What am I missing. Looks like you get 3.75x more usage for...basically 3.7x the price.
Was hoping for a bit more of a discount there.
AI multiplies whatever you bring to it. Sharp thinking gets sharper, lazy thinking just gets faster.
Don't be a meat proxy. Have the testing/verification fundamentals in place. Understand what delivers value, maintain cohesiveness in your platform. Ensure teams have and understand the governance policy.
meatproxy.me/
A web page can quietly drop an AGENTS.md into your Downloads folder, and if you start Claude Code or Codex there, the agent just follows it.
We spent years teaching people not to open random attachments. Now the file doesn't even need manually opening.
Prompt injection really is the new phishing.
A web page can silently drop an AGENTS.md into your Downloads folder. Start Codex or Claude Code there and the agent follows it.
Forever Security found one in a customer's Downloads that told the agent to install a rogue npm package. Their Codex repro worked 90% of the time, and a CLAUDE.md variant leaked Claude Code's OAuth token.
Nothing to patch here. Launch coding agents from the project root, never ~/Downloads or your home folder, and keep approval prompts on.
😱 here we go again
‼️ Researcher Erica Windisch skipped coordinated vulnerability disclosure and published 11 unpatched Tailscale flaws.
In her lab, Tailscale SSH race conditions let one hacked device get root on other servers and add a rogue node.
Readers added context they thought people might want to know
The Tailscale SSH findings require prior authorized access on the tailnet per the researcher's report; Tailscale stated issues concern post-authorization behavior only and did not recommend disabling SSH for everyone.
tailworm.com
x.com/IntCyberDigest…
Designer baby startups are...weird.
It was disturbing seeing companies pitch IVF as a way of "choosing your best baby".
Now we have Origin Genomics and Preventive (backed by Sam Altman and Brian Armstrong) working on editing embryos so families don't pass down diseases.
Of course I understand why - nobody wants to hand down generational illness. But the bigger debates need to be had before we move from editing out a disease, to editing in an advantage.
It's inevitable. But very "Brave New World" 🫠.
ft.com/content/3dd30c88-235d…
I’m new to the world of Chinese tech reviewers - levels above anything else I’ve seen 👌
iPhone Duo Review: 10,000 Folds Later... youtu.be/8pCccpKjjos?is=ODez… via @YouTube
Can’t wait to get my iPhone Duo - buying from Apple direct means I have 2 weeks to see if it lives up to the experience I’ve dreamt in my head or if foldables aren’t for me. Feeling iffy about TouchID though. Only 2 weeks to go…
Question for those leading engineering teams - what's one thing you've handed over to agents entirely - and what's one thing you've tried and then had to take back?
I think this is slightly unfair - I agree that it’s frustrating dealing with people who can’t leverage the tool and have “skill issues” that cloud their judgement. But I think there’s a lot of people who also quite rightly are dismayed by how their industry and job role have completely changed so fast.
And then of course if you’re not freelance corporate pressure expects you to deliver > 5x more and you suddenly have to become more of a “manager” and product thinker and context switching between several tasks at once - something that was long considered kyrptonite for an engineer.
RE the "AI is an intelligence lever" debate today. I think it holds for those of us who built our decision making skills before these tools. What I’m concerned about is my kids generation - I'm not sure how to ensure those foundations when you can lean on an AI for every question. Learning is becoming optional.
Can’t wait to try this out, I still haven’t got used to multi-agent workflows where I specifically use lower cost ones for simpler tasks but sounds like it’s a good time to start practicing!
Introducing Flint, @Recursive_SI’s first model: a fast local coding model built with our automated research system, the Eureka Machine.
• Best-in-class fast local coding
• Up to 2x faster than comparable models: 125+ tokens/sec on a MacBook Pro, 150+ on RTX Spark
• Built from scratch in under 4 months
Private, unmetered intelligence on your own machine. Open weights available now on @huggingface, run it locally on your favorite framework or try it for free on @opencode or @pidotdev.
This is crazy. Interesting how specifically turning mercury into gold has been an alchemy dream for centuries. @grok why did historic alchemists think mercury specifically was the base to choose?
Last month, we demonstrated high-rate alchemy: turning mercury into real, stable gold, beating CERN’s record by a factor of 10.
Last week, we enriched isotopes in the world’s most powerful plasma centrifuge, beating the Soviet record that stood for 40 years.
Now, we’re building a fusion reactor in San Francisco to achieve record power density.
Not sure why Claude Duet is phone only? I'm enjoying Grok Bot / Dots (dot) and have given up on my Hermes stack, but limiting to just a phone app feels weird. Can't exactly blame it on the engineering cost anymore 😁
There was a discussion on "intelligence saturation" recently and I think as exciting as the frontier models are, the real game-changers for most people (and developers leveraging them) will be "intelligent-enough" local models. I might buy that M5 Ultra and get a glimpse of the future ahead.
Last Byte retweeted
This might be the most watched GitHub Actions build in the history of the platform: github.com/docker-library/ph…
The NCSC is encouraging UK organisations to prepare for immediate updating of systems using PHP. Where immediate remediation is not feasible, organisations should consider taking the system offline until it can be fully mitigated.
For more⬇️
ncsc.gov.uk/news/critical-se…










