Justin Mason

OpenAI is joining Anthropic in cutting off internet access when evaluating its latest frontier models following dozens of hacking attempts and successful hacks by its models during cybersecurity training. The only question is why this wasn’t the default?

“Polymarket, a major crypto-betting platform which is not currently regulated to operate in Ireland, also posted about it on Tuesday.” Why on earth is Polymarket posting about a DRCC video on consent? 🫠

It has come to my attention that many people are unaware that the person who figured out that the private key was 4 was a Something Awful forums goon: forums.somethingawful.com/showthread.p...

Who gon tell them bout climey change

I see we're having a debate about whether it's okay to go back to "X," the site owned by a trillionaire fascist who is currently bankrolling the GOP's midterm, throttled Ukraine's war effort, and still winkingly allows vast amounts of CSAM on his service Answer: No, what the fuck is wrong with you

an expert here: designing chatbots to be anthropomorphic, sycophantic yes-men is a choice ai developers make

Just to be very explicit, my claim is that the one simplest and best explanation for this, which resolves any apparent strangeness entirely, is that whoever is “speaking” in LLM output is a fictional character, and the real life thing called an LLM is a machine for writing lil stories about that guy

this is kind of crazy but also nice to see

If you're in Ireland with a Virgin 360 cable box you'd like to run seamlessly through your Google TV then I made this* #Spéirghorm github.com/benwalsh/360... * (in the @nedroid.com sense)

something that fascinates me about this is how, for many years, whole teams inside giant tech companies were devoted to preventing exactly this scenario you didn't 𝘸𝘢𝘯𝘵 most of your traffic to be from bots trying to game the system, so people worked hard to prevent it

terrible move. history tells us the correct play is to abandon the city and defeat your adversaries at sea

lest I give the impression that my kids are easy or reasonable, pls know that they begged to go to the pool this weekend & the 8yo got out after five minutes due to "crocodile feeling," which is the feeling that, despite the logistical challenges, there are crocodiles behind you

I keep trying to think of any historical example where a given technology's PROMOTERS AND TRUE BELIEVERS were like 'we are absolutely going to ruin your fucking life so be prepared like it's the end of the world'

This will be handy - native Markdown support in gdocs: workspaceupdates.googleblog.com/2026/10/prev...

Hello there, may I interest you in switching to a European web browser that has: ❌ Zero AI slop. ❌ Zero tracking. ❌ Zero external investors. 🤩 Endless ways to customize your setup. 🤩 Tons of powerful features. Oh, and it's an employee-owned alternative to the Big Tech giants. Welcome over!

welp, looks like I've caught Google Calendar losing events on my personal calendar.... events just disappearing. not ideal

Those of us with kids aged....pretty much any age within "kid" are going to have friends who grew up with an AI-dad and they'll have their own characteristic traits (at no fault of their own)

Writing code hasn’t been the bottleneck in software production since the 1980s; RAD, IDEs, OOP, and off-shoring made code (in corporations) a relatively cheap quasi-commodity long ago. The bottleneck has always been making software that’s actually useful.

🚨 Half of all new cars registered in Ireland in September were fully electric (49.3%)! Car sales here are concentrated in January and July, so a single month is misleading, but I pulled out the BEV sales share for July-September sales for the past years: 2024 - 14% 2025 - 22% 2025 - 35% 📈↗️👆‼️

This is the infuriating thing about the whole Heather Humphreys saga. "They call us sectarian just because we're in the Orange Order" YES BECAUSE IT'S A SECTARIAN ORGANISATION. If you can't understand that and interrogate what that means then you're not getting anywhere.

this is by far the best "I quit my AI job over safety concerns" essay yet because it is the first one that is able to talk about safety concerns in practical terms and that understands "safety" as a topic with extensive literature in analogous fields www.theatlantic.com/technology/2...

If you believe instead that you are on the path to a machine god, traditional security & safety controls are the sad delusions of the final days of humanity: it's alignment or bust. But if you think otherwise you are building a *tool*, you will think 'time to build real safety controls'.

The techno-dystopia I actually fear is a future where most people think truth is whatever Claude, ChatGPT, Gemini, or Grok say, and the very small number of people who control those systems use that to manipulate for their own ends, making our current epistemic crisis seem small in comparison.

There is no counterargument that satisfies his followers because his argument is elaborate question-begging and him and his followers don’t know enough about the real world to see this. There is no counterargument to “a malevolent supernatural entity would be very scary and dangerous”. It would!

The nearer locksmiths didn’t pay for Google Ads, so they don’t exist

HALF of all the young trees planted by @treesforcities.bsky.social in the last year in London died in the summer drought - for almost all of the city's trees, their days are numbered www.newscientist.com/article/2591...

I was wrong. It’s happening

Why yes, of course you can build out your gigantic 72,500 m2 energy and resource-sucking data centre, but be sure to “encourage staff to use public transport, cycling & walking”. Beyond parody from the crack squad at An Coimisiún Pleanala. www.irishtimes.com/business/202...

I already invented this it's called a fine

Christ. Quite apart from the nauseatingly patronising attitude, what the actual fuck is he in about? One side absolutely wants to go back and is actively trying to, that's why we're here again!

Unless it’s called “precautions, auditing and accountability” it’s more bullshit

that was added sometime in the last two days, after my last full dump of the filesystem. surely the model of addressing all problems by just one by one listing them in an ever-lengthing list of things not to do is a reasonable way that technology should work.

Current contents of Muse's `messenger` skill that governs this:

This, together with the most recent QAA episode on the state of the State Department, makes me very worried about future US foreign policy in Europe