Do you think that Europe should bargain away its digital sovereignty to appease Trump and the broligarchy? Strong majorities in Germany, France, and Spain are against that (YouGov). @coricrider.com and I have a better plan: www.politico.eu/article/digi...

it feels important to keep saying: SPA as an app framework default (and not an opt-in feature) was a mistake

Enjoying building tools from scratch in an effort to reshape how the digital tech industry develops. rai.onl is another pillar forming a set of foundational tools that I hope provides more responsible alternatives in the future.

Google threatens to pull apps from Play if they don’t receive updates within 12 months. Building a stable service and having a feature-complete app is fundamentally shaped to be a bad thing. Yes, I just had an account with several apps closed. They won’t reopen it once they deactivate.

In September I’ll be giving a special talk at the @mydataorg.bsky.social conference on zero-compromise privacy-preserving AI, using private data spaces, able to run anywhere. No cloud, no surveillance, no theft.

Bluesky CEO @jay.bsky.team is right, we need to stop thinking of social media as a zero-sum game that platforms are “winning” or “losing.” Why not measure friendships, or genuine connection? We need to rethink what we value in our online tools for socializing.

Calling the US IRS because a form cannot be submitted electronically. Options for a first-world country: telephone, fax, or letter. Reading the form out over the phone because capitalism is clearly a winning strategy. 🤦

One of my publishers has asked me to write a book on vibe coding. … entering treacherous waters.

Is there an application that can stream to a destination securely and with privacy intact so that if a phone is suddenly taken / destroyed the last part of the feed will be retained safely? Preferably open source for trust. If not it’ll be today’s form of procrastination.

If you're a threat actor and try to use ChatGPT or Claude for operations (cyber/info), you clearly don't get it. You're proactively setting yourself for a failure. I can't imagine any serious cyber threat actor doing it this way.

Just ended a call with a well known pharmaceutical conglomerate who want help building AI monitoring for their production lines. It’s 2025. How does a large pharmaceutical company not have this already in place?

I’ve got a new project goal: finding something where I can work with @set.studio.

We've got availability at Set Studio this summer! We're wrapping up some great work which we can't wait to show you, but for now check out some of the stuff we've achieved 👇 If you've got a project idea cooking, we'd love to hear it. set.studio or hit me up with [email protected]

Thanks to everyone sharing this! 💕 On an off chance someone wants to give a mid-career woman a shot at browser engineering, I’d be very very open to that. I’m new to it but I’m working on contributing Close Watcher to Servo atm & learning pretty quickly.

I still use maps, read road signs, and pre-plan routes and I wasn’t even old enough to drive before GPS. It just seems logical to know what you’re doing before setting off. Also, I love maps.

Instead of Cartier diamonds, invest in Kyber / CRYSTALS.

New post on the Swift blog: we re-wrote the Password Monitoring service for Apple Passwords in Swift and saw huge improvements in memory use and throughput. Some of the details here still blow my mind; it's a fun read. https://www.swift.org/blog/swift-at-apple-migrating-the-password-monitoring-servi

This is exactly what’s happening, except it’s not best practices that are being propagated. It’s the most common practices, which are rarely the best.

Digital sovereignty is almost entirely an infrastructure problem, and it's a tough one. How can we understand the problem and start outlining strategies? Read on to find out! berjon.com/digital-sove...

happy pride month, y'all! 🏳️‍⚧️🏳️‍🌈

Job hunting is a truly soul-sucking experience, so I'm posting my CV here just incase one of you needs a technical leader with web standards, dev advocacy, or strategy experience. Please save me from the 4th circle of hell (aka Linkedin) 🙏🏾 You can see my CV in full: docs.google.com/document/d/1...

Incredible progress yesterday. I closed 795 tabs in one go and I have no idea what was on them. Slight chest tightening, but otherwise good progress. If only that was the only browser window I have open on that device… or the only device I have.

More or less that. I think we are looking at an unprecedented form of destruction of knowledge through the creation of infinite content.

Based on these discussions it’s unsurprising that folk think AI models are only LLMs, and only big companies build models from scratch. Last week I wrote two models in Rust specifically to handle healthy recipe generation. They were just over 20MiB in size, and can run easily on Raspberry Pi Zeros.

how many had banned generative AI in their organizations? Three hands went up. "And how many, in your deepest of hearts, actually have a good grasp of the security risks involved in AI system controls, by a show of hands?" Not a single hand was raised among the 200-strong, security-savvy crowd.

“Without widespread theft, my industry can’t survive” is an argument against your industry surviving, and is not an argument for allowing widespread theft

welcome to the future, now your error-prone software can call the cops (this is an Anthropic employee talking about Claude Opus 4)

Folk getting excited by, and celebrating, annual conferences by US brands showcasing more proprietary tech usurping freedoms of the open web is one of the most convincing demonstrations that the digital community finds it hard to learn from previous—and ongoing—issues.

We’ve been building an alternative sovereign mail service for years at @omnifi.foundation. It comes with incredible hurdles. I’m working on revisiting this work and launching a real alternative (self hosted or / and portable) as part of @naamio.cloud.

I just wanna add to this how much hope you're injecting into Piccalilli/Set Studio HQ. We need a big win in this industry of ours. The wall-to-wall slop and KPI chasing is bringing us all down. Your support is what's going to make this happen too.

Each time you ask an AI chatbot a question, it sends a request to a data center and strains an increasingly scarce resource: water. A 🧵on our investigation into how the data centers that power AI are increasingly being built in highly water-stressed places: www.bloomberg.com/graphics/202...

I can’t believe I have to say this, but please don’t use a chatbot as your therapist www.theverge.com/policy/66568...

There was a moment when explainable AI (XAI) was going to be regulated, forcing every LLM to disclose the source. OpenAI said this wasn’t possible (which was always nonsense). @ai2.bsky.social introduced a novel technique to showcase the bare minimum of what we should expect from language models.

A little known fact, the @yle.naamio.social feed, from the English-language, Finnish national news was built by me over a year ago in a couple of hours, including thoroughly testing and deploying. It was intended as a short term experiment, but it’s been running for over a year.

Last week I wanted to explore optimizing federated learning and PETs in Rust. The ecosystem isn’t as mature as Python, but the benefits of building privacy-preserving AI in Rust over Python are significant. For two days I wrote two implementations. One with Candle, one with Burn.

Hyperlight is super interesting for this space and goes along way to resolving some of the work we did last year at @omnifi.foundation to provide safe AI federated learning enclosures on the edge through custom WebAssembly hosts.

Received a request to help a company remove its software engineers from product development, and instead teach the sales team to use AI to build new SaaS products—from scratch, and into production—in 2-weeks based on what customers directly ask for. It’s getting hard to wade through the nonsense.

Alright, we're actually doing this open working thing. We're going to work on projects, for real clients who are set up to make people's lives tangibly better. We can only do that if it works financially though because we've got mortgages etc. Your support will make this happen.

Folk seem to have taken my comment that we should all be luddites as an attack on technology progression, which is a rather ignorant view of the Luddite movement. It was a rebellion against the impact of machinery on workers, including fair pay, and on the low quality output. Be a Luddite.

Zero patience for folk using whataboutism for the negative impact of AI slop. It’s lazy and is about shifting responsibility. Parallels exist, sure. Take the Industrial Revolution. We should all be luddites.

Happy Europe Day! 👋 It’s been an interesting year so far, as we see a significant push to building in sovereign European data centers. We’re working with a number of sustainability-focused compute providers to make building in Europe as easy as building in the US, (and more cost-effective).

Almost 20-years of building AI-based technologies and I am all for the term “AI slop”. Being able to differentiate from useful AI, (disease prevention, road safety, food supply, etc.) and mediocre, snake oil-fueled drivel, helps reduce some of the harm the AI field is suffering.

I genuinely don't understand the "using chatGPT isn't an energy hog" statements. You are right in a very narrow way, yes. If people believe that simple use of AI output is a massive energy hog they are misinformed. But you're dodging the ongoing massive cost of training, tuning, etc.

The @wasm.io talk I did this year has had some of the most constructive tail discussions of any previous talk. There’s a lot of energy being spent on genuinely new ideas where the technology presented is a key enabler.

Computer science is definitely political. From the tools you pick, to the solutions you build, it’s political. I’ve been meaning to write something up on this. Far too many engineers and data scientists seem to feel they’re politically agnostic. That position alone is a political choice.

Invited for coffee to be asked about taking on a CTO role for a video game tech company. Interesting premise using Rust and WebAssembly to improve the speed for developers to build. Then they mentioned a16z was involved and everything else became white noise.

Everything @rudyfraser.com and the community is doing with @blacksky.app is amazing. Checking in on the repositories weekly and seeing how it’s moving is inspiring and this is fully deserved.

Strava (and other run-tracking services) have had a terrible time with regards to privacy. In some cases lives have been lost due to location leaks. It’s great to see things built on ATP but safety-by-design is absolutely critical—and currently missing—when building services on a public graph.

Disappointed I couldn’t make it to @ahoy.eu. Buoyed by the posts though. Trying to build things when constantly moving is challenging, but it’s encouraging seeing so much positive engagement in the community. Maybe next year, if I can stand still for a moment.

Good technologists have to accept nuances. All technologies are flawed to some extent—especially digital—and LLMs are more flawed than most. Blankly disregarding their usefulness outright is simply poor critical thinking. In many cases other forms of AI are far more effective than LLMs though.