If every LA decides to procure 5 new API-less monstrosities it will be catastrophic. Potentially even more so if they decide to vibe-code them up themselves or get taken in by Shiny AI Shysters.

The website still says 14. One of the service members not being counted is Sgt. Angel Sarah Rampersad from Queens. She was married to a woman she met in the Army, and one of our podcast listeners pointed out that fact was not acknowledged at a recent memorial service. So I am acknowledging it here.

Charlotte Clymer@charlotteclymer.bsky.social · 2w ago

As of this morning, Trump's Pentagon still only counts 14 service members as having been killed in the War in Iran. The Pentagon claimed last night it was a tech error. It has not been fixed. They are refusing to count four service members. You can see it here: dcas.dmdc.osd.mil/dcas/conflic...

Any incoming minister for AI should be tasked with replacing a load-bearing spreadsheet from a critical service, to give them the insights into why no one has done it yet. The ministerial equivalent of breaking a horse.

These heatwaves show leadership cannot mean stepping back from climate action - important piece from @alistrathern.bsky.social www.politicshome.com/opinion/arti...

These heatwaves show leadership cannot mean stepping back from climate action

The UK is in the throes of our second significant heatwave of the year. It is still early summer. That fact alone ought to give us pause.

politicshome.com

"It looks like you're trying to send an attachment but haven't attached" is super helpful. Please please please can we also have "It looks like you're booking in a catch up with Alice but you haven't add her to the invite" I'm begging

Love this as a prevention case study: - Low cost, even revenue raising in the short term - Improves health outcomes and reduces costly hospital admissions Shows political leadership is key: - Politically costly for Khan to implement - Did it despite cost savings landing outside GLA

The Guardian@theguardian.com · 2mo ago

Emergency hospital admissions fell after introduction of London’s T-charge and Ulez, study suggests

I've spent a lot of time thinking about this conundrum and, about a decade ago, ran a small cross-party programme on tech literacy for MPs. A quick 🧵 about why MPs know so little about tech.

Ann Kempster@annkempster.com · 2mo ago

The absolute lack of understanding of how tech works by Government ministers is embarrassing and frankly, in 2026 100% unacceptable and inexcusable. Barking.

Another write up of the Microsoft press release with no mention of whether use of personal data will be allowed. Without this, these licenses are pretty useless.

The Register@theregister.com · 2mo ago

NHS prescribes half a million Copilot licenses for its paperwork headache

Unless the governance has been changed to allow patient data to be used in Copilot (it wasn't in the trial) then this ain't going to be delivering what most people might think reading the Microsoft press release.

Digital Health@digitalhealth2.bsky.social · 2mo ago

NHS England is rolling out Microsoft 365 Copilot to 505,000 clinicians and support staff across healthcare services. www.digitalhealth.net/2026/06/nhse...

It's not good for the country if we're constantly junking PMs (of any political stripe) because of the vibes. So much reporting is now vibes, not properly looking at policy, how that's delivered etc. That does everyone a disservice.

it's also, unlike a lot of the other issues that are being pored over, very relevant to local government. It's unclear what levers hackney council has to pull on gaza, but there's a hell of a lot of ways they can make life easier or harder for trans people in the borough

What's different now is the blast radius. A botched microservices migration costs you a re-architecture project. A botched agent deployment can apparently delete your business in 9 seconds. 4/4

Politicians just seem to have given up trying to explain the world as it is to people. It's a huge problem.

Stephen Bush@stephenkb.bsky.social · 3mo ago

The fuel shortage has left me depressed about British politics in a way I never have been before, in that not a single party has avoided either saying or doing something incredibly stupid and completely removed from how anything actually works.

The first independent review of Mythos cyber capabilities I've seen. The UK's AI Security Institute - an independent government body - appears to have got their hands on Mythos and published their findings. 1/7 🧵

Our evaluation of Claude Mythos Preview’s cyber capabilities | AISI Work

We conducted cyber evaluations of Anthropic’s Claude Mythos Preview and found continued improvement in capture-the-flag (CTF) challenges and significant improvement on multi-step cyber-attack simulati...

aisi.gov.uk

Claude Mythos is that guy down the pub who is so good at karate that if he used it on you, you'd die instantly, and that's why you'll never see him using karate.

There's a computer science term for what OpenAI are doing here, we call it "lying"

Ciaran Martin@ciaranm.bsky.social · 4mo ago

Companies have been blaming AI for cutbacks which are really caused by economic conditions & weaknesses in their business model. Now we have an AI giant blaming wider economic & regulatory conditions for a cutback actually caused by weaknesses in its own business model www.bbc.com/news/article...

Putting aside Sam Altman just continuing to emit huge lies constantly, this is an amazing example of what I mean when I talk about how GenAI is replacing *extremely low energy* digital tasks with *WAY WAY HIGHER ENERGY* replacements that *do not even work* Why TF is anyone using ChatGPT as a timer!

Sam Altman Says It'll Take Another Year Before ChatGPT Can Start a Timer

An $852 billion company, ladies and gentlemen.

gizmodo.com

BREAKING: After ProPublica published letters written by children at the Dilley concentration camp, ICE officers raided the dormitories and confiscated the children's letters. I'm suing ICE to get them back. My latest: www.muellershewrote.com/p/breaking-i...

NEW: ICE Raided Children's Dorms at Dilley and Confiscated Their Letters. I'm Suing to Get Them Back.

In our latest lawsuit, MSW Media is demanding children's letters confiscated by ICE agents, and the Office of General Counsel memo authorizing warrantless home entry.

muellershewrote.com

1. LLMs do not reason 2. LLMs do not understand anything, they have no concepts: "pressure", "morals", you name it 3. It's a great headline but to "threaten" to use nuclear weapons is one thing (and actually key to MAD) but article also says LLMs going "full-scale nuclear war was rare."

King's College London@kingscollegelondon.bsky.social · 5mo ago

AI bots threatened nuclear strikes in 95% of simulated wargames. A study by Prof Kenneth Payne, School of Security Studies, examines how these systems reason under pressure - as militaries & security institutions increasingly experiment with AI for analysis & planning. www.kcl.ac.uk/news/artific...

It would be a very very very good thing for Starmer to show some principled leadership around now. And perhaps it might also focus the mind a bit, away from playground disputes.

I'm astonished this statement could be reported so uncritically. Let's be real: *IF* AGI is ever developed, all its efforts will be spent making OpenAI profitable and sending Musk to Mars. From www.theguardian.com/world/2026/f...

That over-reliance on a heavily-subsidised technology isn't seen as a strategic risk by engineering leaders is baffling.

Before you start attributing this to AI, note that the biggest drops in employees has been in retail and hospitality - sectors with 122,000 fewer workers than this time last year. (we are kicking off a project on AI workforce impacts though, so watch this space...)

"[The] report reveals three-quarters of CIOs admit to not having full real-time visibility into AI agents running in production systems, even though they have the ability to implement actions." Only a matter of time til agent goes wrong and there will be panic. www.theregister.com/2026/02/17/n...

CIOs told: Prove your AI pays off – or pay the price

: Boards demand measurable ROI as budgets, bonuses, and jobs hang in the balance

theregister.com

Really really wish Oracle were not calling this an "agent" which suggests an autonomous AI tool making decisions without human in the loop (which this almost certainly is not, because that would be nuts)

Digital Health@digitalhealth2.bsky.social · 6mo ago

US tech giant Oracle Health has announced the launch of its clinical AI agent in the UK following a NHS pilot programme. www.digitalhealth.net/2026/02/orac...

The business model of LLM makers is blatantly unsustainable. In any other domain these companies would have hit the wall years ago. Anyone using LLMs for anything important needs to be asking "what happens when we have to start paying 10x current price for tokens?"

Ed Zitron@edzitron.com · 6mo ago

I fundamentally believe this era is an illusion. The current versions of Codex and Claude Code are costing OpenAI and Anthropic so much money because they’re running at like a negative 150% gross margin. All improvements involve more and more compute. The limits will be sudden and brutal.

Major UK research resources get long-awaited access to GP data Naomi Allen, chief scientist at @ukbiobank.bsky.social tells me it is a “watershed moment” for the resource and for global health research in @resprofnews.bsky.social www.researchprofessionalnews.com/rr-news-uk-i...

Major UK research resources get long-awaited access to GP data - Research Professional News

Government directive for sharing health records hailed as “watershed moment” for projects like UK Biobank

researchprofessionalnews.com

I wrote an article for Ada National College for Digital Skills, who run digital technology apprenticeships in the UK

GenAI is amplifying the skills gap in software engineering

All the available evidence suggests that GenAI-assisted coding is most powerful in the hands of highly experienced software engineers, while having neutral or even negative effects for less experience...

linkedin.com

I don't think people appreciate enough the fact that, at a time where global warming was approaching an inflection point yet renewable energy was finally becoming economically viable, we decided to squander that opportunity because literal death cultists decided we needed Infinite Bullshit Machines.

The next round in our game of Who Wants To Be A Software Developer is called "Patching and Upgrades". You'll love it! www.theregister.com/2026/01/30/o...

OpenAI axes ChatGPT models with just two weeks' warning

: GPT-4o gets second death sentence after last year's reprieve, but this time barely anyone's bothered

theregister.com