Reader, I read the sign like it was a movie trailer. “Coming soon…”

If you're new here and happen to see this, here's a Critical Tech starter pack, full of interesting folk to follow go.bsky.app/VFxSES6 And if you have a small account (<10k followers) and would like to be added for visibility, let me know! (Although response times may be slow)

Fiercely proud of this team. For their work, care and impact. medium.com/writing-by-i... @projectsbyif.bsky.social

Got some iconic swag for IF today. My dream is to eventually make something like this uniform from the ICO enforcement “raid” at Facebook in 2012. Badass. The FOI request “Will they be made available to the public for purchase?” 🔥 www.whatdotheyknow.com/request/ico_...

In 2016, godfather of AI Geoffrey Hinton said: "People should stop training radiologists now." In this week's The AI Shift newsletter, we ask: why are there still so many radiologists?? www.ft.com/content/f2e0...

IF is *almost* 10 years old. Wild. I never really thought about a decade of work - yet here we are. Funny thing: the work feels more important now than when we started. Building muscle memory for how orgs handle power, automation, uncertainty. Still worth doing. Still worth doing well.

My talk on moving fast, safely with AI is live youtu.be/f11Otsh6OKs?... thanks to everyone who came to listen in person and to our wonderful panellists afterwards!

"I love my job…And even though my job requires me to commit sociopathic acts of evil that directly contribute to making the world a measurably worse place from Mon through Fri, 5 days a week, from morning to night, outside work, I’m actually a really good person” www.mcsweeneys.net/articles/i-w...

inspired by this, I have done a free extract of this section of the book here. it is about understanding that *with all its problems* the BBC is as important for our intellectual health as the NHS is for our physical health: naomialderman.substack.com/p/the-bbc-an...

Every organisation now faces a choice: experiment carefully or fall behind the AI curve. Learning safely is key: teams that understand the risks build systems that earn trust and drive adoption. More on this in our latest newsletter from @sarahtgold.bsky.social www.linkedin.com/pulse/12-bui...

Password1234

Excited to have Christine Foster, General Manager GenAI and Automation at Experian joining sarahtgold.bsky.social on the panel at our event 'Crossing the AI divide' We'll be discussing how ​Experian are introducing AI into their products & consumer experiences - without breaking trust

Petition to increase funding to UK cooperatives as advocates for community building and peace: petition.parliament.uk/petitions/72... Thought it’d be of interest to many of you here! ✍🏼

In other news… an invitation The AI gap is widening, join us for an evening of real talk about AI. Not the hype, the hard part. On Thu 13 Nov, a talk from me, & panellists from Experian, DSIT, Monzo & DeepMind, chaired through a discussion by Richard Allan. ✨ luma.com/xh4nnffz

Oh so shattered travelling to New York right now, but oh so brilliant to spend a week in person with a client. To get in a room, talk, align, break bread, dream. Time in person is just so damn good.

Today’s the day I remember why I don’t have a smart home

A good explainer on the enshittification of much of the internet … IMO modern digital services in the public sector look increasingly radical when they’re built for clarity, trust, and transparency (as they should be) www.theguardian.com/technology/2...

On Friday my phone blew up with messages about BritCard. I don’t think it’s a tech test. It’s a trust test. 2m+ signatures against already. If trust isn’t designed in from day one, BritCard risks repeating the same failures that killed past ID schemes. My take medium.com/writing-by-i...

Remember kids, each of these can be made easier with an ID scheme but a) it doesn't need be same ID scheme, b) most of effort is in rest of system and services around ID scheme (eg getting nurseries to use same processes/systems), and c) some of these are *already happening in UK*

The head of the UK's leading anti-racism campaign group Hope not Hate wrote to Keir Starmer to complain about his "silence" in the face of a rising far-right threat. "At the time when people needed leadership there was silence. “And for the far right, that silence just emboldened them further"

City hall looking lovely this afternoon. But I’m running late so no time to enjoy the sunshine, alas!

“Fail fast” is a common mantra in tech. But when it comes to AI, success isn’t about how fast we experiment. It’s about whether people actually use what we build. We're hosting a panel event on 13 November in London on building trustworthy AI. 🔗 Register for your ticket luma.com/xh4nnffz

Reform hold a press conference every Monday, for which they trail a policy announcement in that morning’s papers. Given Monday is a slow news day, that often means they lead the papers – and so the breakfast shows, and so they set the agenda for the week. It’s a good trick, but easy to counter!

Totally agree. Online, Reform's announcement is are above the fold on every news site I visit. Why do we give them this coverage?

Working from a new coffee shop in my local library. 10/10.

The fundamental challenge in technology ethics is what you consider as the boundaries of the system. bsky.app/profile/duhe...

Enjoyed listening to this episode with @timbl.bsky.social interviewed about life, obviously the web, and what he thinks about where we’ve landed. I like the idea that podcasts are similar to the original blogosphere. Not thought about it like that before.

Coming back to work on Monday. If you run a human-centred, people first business like me, and you can use your platforms, I invite you to say something about the alt-right march on Saturday. We need to practise what we preach. (& soz for screen shot from LinkedIn…)

I've written to Keir Starmer, Kemi Badenoch and Nigel Farage, urging them to join me in condemning Elon Musk's dangerous remarks inciting violence yesterday. As leaders, we must stand together and make clear Musk will face serious consequences for these actions.

I became a trustee of Open City recently, so really nice to take part in Open House London to see “where the magic happens” inside some of my local buildings. Today we visited Kaymet - a manufacturer that’s been hand making beautiful trays since 1947

In other news - if you’ve not seen Do Ho Suh exhibition it is beautiful! The details are something else

Chose the wrong day to visit the Tate. Far right march is congregating, ready to start. Flags with “Close the borders! Protect our children” printed on them. So many people. It’s making me feel dread, and fear.

There needs to be space created by AI and OS devs so that there are ways for apps that people use for privacy to be able to make privacy guarantees that aren't immediately undermined by an AI agent that has been granted the ability to do literally anything on their behalf. Headline is trash though.

Mentoring works. "The most significant thing that has changed within me has been my confidence [...] and also my networking and connections have grown. I couldn’t be anymore happy with my mentor." - Current mentee Become an Arts Emergency mentor 👉 www.arts-emergency.o... Apps close 29 Sep 25

New design patterns klaxon! AI isn’t magic. If we automate support without designing for trust, we’ll just repeat the same failures people already hate. How do you make automation trustworthy? 3 new patterns & post from my colleague Ronan medium.com/writing-by-i...

**Jobs** Dear Bluesky folks, we are looking for more folks to join us at IF. Product manager with AI expertise ML engineer Design engineer If you’re interested in working at the intersection of genAI and trust, please get in touch with me & help me share! A mix of perm and freelance

Come and join us to work on some VERY interesting projects that will change the lives of millions of people around the world, with a great team, who like to work hard and have fun too tally.so/r/mZXV6e

AI for public good report highlights that in administrative justice, AI’s value lies in supporting appeals, feedback loops, and fairness, not replacing human scrutiny. Sandboxes, statutory duty alignment legitimacy checks = core building blocks. AI must earn our trust, not assume it.

“The world that comes next will not be the sum of our hopes but the record of our activism” Such a fabulous talk that you need to listen to www.youtube.com/watch?v=-VEW... Only collective action will fix the world. The children in Gaza, Congo, Sudan don't need our hope they need our help.

Our latest newsletter: What changes when you build with AI? When AI meets healthcare, finance and the public sector, cracks show fast. Oversight is costly, delivery rhythms shift, and accountability gets emotional. projectsbyif-newsletter.beehiiv.com/p/9-what-cha...

"AI has become the infrastructure we build around not because it was integrated into our national systems, but because nothing else has been. It has filled the space left open by under-resourced schools, inaccessible mentorship and uneven digital policy" www.linkedin.com/pulse/kenyas...

Great from @carlyayres.com carly.substack.com/p/vibe-codin... Made me think, what if we simplified the language? Not misuse/ misalignment/ edge case If a system harms people, it’s not just a vibe gone wrong, it’s something we should call what it is. A crime. Digital needs accountability.

I’ll be there with Astrid - join us. If you’re also parenting today, come with your bambino and a pot. Let’s let out some rage and show our kids what democracy involves.

The UK govt is now working with OpenAI, Google DeepMind & Anthropic to “find use cases” for foundation models in public services. It’s framed as exploratory, a test of productivity + citizen-facing improvements. But these partnerships go deeper than tools. 🧵

The Daily Express front page today, 23 July 2025 It does indeed shame us all.

I'm doing some work on feedback in public services (that includes AI stuff), kinda like this medium.com/writing-by-i... Is there anyone out there who follows me who's also working on that? Or any of the existing feedback channels (like the gov.uk "is something wrong with this page?" pattern)?

Another week, another AI trust crisis. This time: WeTransfer. Reports implied user files were being used to train AI. The company denies it, but just like Dropbox last year, the damage may already be done. Trust erodes fast. Even when the facts don’t back it up.

Never thought I'd be talking about tech, trust and tennis...but here we are! I've finally read up on the Hawk-Eye failure at Wimbledon. And I think it reveals something more nuanced and useful than the big headlines make out. 🧵

A great looking role at IF and a really nicely framed job description that looks like it does a good job of explaining the org culture as well as the role.

The 424th edition of Internet of Public Service Jobs is out - including a spectacular role with @projectsbyif.bsky.social newsletter.digitalbydefault.jobs/internet-of-...