There is a massive push for AI from the Norwegian government, but yet again, as happened in the 90s, the focus is on using and consuming, not creating or producing. We make ourselves dependent on American tech companies yet again.

«Researchers worry that overdelegating decisions to AI could gradually weaken the critical thinking and judgment skills that underpin independent thought.»

The history of Lorem Ipsum

time to assemble and test #meshtastic @patigallardo.bsky.social

Before this LLM dark age we’re entering, my view was that we needed more social sciences in software and hardware development. Now here we are. Quality, usability, security and other «non-functional», qualitative requirements are out the window. Lines of code is suddenly a valid metric of work.

I think QAs have a bright future. These people are generating whole codebases they don’t even understand, massive programs with an unknown number of bugs. They’ll need a whole army of QAs just to get an overview of the state of the system.

We **unlearned** best practices and wisdom we accumulated over **decades**. We threw it all out the window in favor of speed, counted as lines of code per hour. We have books from the 70s and 80s explaining why that is a terrible measure. But we forgot that too.

We’re just gonna do years of this aren’t we?

This means that your tests are the only thing of value. Everything else can be regenerated. So, how’s your test coverage? Where’s your QA?

Always good to know who your opponents are

Someone mentioned Gell-Mann Amnesia for AI generated code, and I’ve been saying this irl, but thought I was imagining things. I am 10x more impressed with AI generated code in languages I’m bad at vs languages I’m pretty good at. en.wiktionary.org/wiki/Gell-Ma...

It’s fascinating to see the lack of cognitive dissonance when programmers that wouldn’t use a vibe coded library, think *their* vibe code is Different Actually.

I’m not interested in paying for bad information or bad code. This is undermining the entire value proposition of science. This is a vibe based dark age we’re hailing in

This collective LLM delusion is completely baffling to me. People are talking like truth, correctness, precision, performance, security… none of it matters anymore. Generating a lie is «good enough». Wtf is wrong with people? If correctness doesn’t matter, why would I pay you? A wrong answer is free

Most of these AI evangelists sound like Agile Coaches.

Do they hate programming? What makes them think they are qualified to guide this when they can’t even be bothered to look at it?

I’m watching programmers basically fuzzing themselves to an implementation. Write lots of tests, or specs, then have an LLM generate stuff and test it against the spec, iterate to infinity until it’s «good enough». The implementation is black box. Jesus, our field has digressed decades.

This collective LLM delusion is completely baffling to me. People are talking like truth, correctness, precision, performance, security… none of it matters anymore. Generating a lie is «good enough». Wtf is wrong with people? If correctness doesn’t matter, why would I pay you? A wrong answer is free

Research confirms that the more skilled professionals like doctors and software engineers rely on AI, the more their skills degrade. www.nature.com/articles/d41...

If lawmakers are serious about protecting children, they should roll back age-verification policies and start targeting the systems that incentivise mass surveillance. Read my latest print piece for The Guardian 👇 www.theguardian.com/commentisfre...

If you want to immigrate to the USA you have to prove that you’re vaccinated against measles. Just a thought that occurred to me.

A reminder that the Voting Village Call for Presentations is open. If you've got interesting work on election security, please submit it! August 7-9 at DEFCON in Las Vegas. votingvillage.info/defcon-2026/...

Happy Juneteenth! When we celebrate the time that racist Texans tried to keep slavery going for another 2 years after abolition. But "the news🤡" of abolition reached their slaves! By "the news" I of course mean Black soldiers from the United States Colored Troops that liberated the Black Texans.

NEW poll shows a Democratic socialist born in 1997 leading against a veteran Democrat who first took office in 1997. The CO-1 Democratic primary poll shows 29-year-old Melat Kiros ahead of Diana DeGette, 41-36. *38-point-swing* toward Kiros since March.

I now use Amazon like I used to use bookstores (remember bookstores?) - browse for what you want. Then buy it somewhere else online once you have the manufacturer, model etc information. A little extra work, but worth it.

"If only they had learned the truth earlier! Someone must carry the truth!🤡" That's... not the lesson of Juneteenth. Get some Black people to read your posts before posting. The lesson of Juneteenth is that racist Texans will try to hold onto the worst racism even two years after it's abolished.

On the state of the Russo-Ukrainian war — very sound and in accord with what I see. By @anneapplebaum.wsocial.eu www.theatlantic.com/ideas/2026/0...

I worry that wounded pride and «who won» games will result in even more civilian wounded and dead. This war should never have been started.

Jfc, Hilary, this reads as if the Palestinian people are performing a genocide on themselves and like the US government hasn’t, across both Republican and Democratic administrations, actively supported this systematic eradication of a people by the Israeli government. archive.ph/2026.06.19-0...

Norway is upgrading its presence in #Greenland by establishing a Consulate General in Nuuk. “Greenland is a close partner. A Consulate General in Nuuk will strengthen political contacts and cooperation on shared interests” says PM Jonas Gahr Støre. 🇳🇴🇬🇱🇩🇰 #Nordics www.regjeringen.no/en/whats-new...

I'd be delighted, frankly, if they proved me wrong here. As someone who as warned of vulnerabilities in US election systems, I would LOVE to see actual evidence that these (real) weaknesses have been exploited. But so far, crickets.

Some of those sockpuppets are obviously bot accounts - they post only shrill defenses of ETA, and have no interest in anything else. Others are naive true believers who've been hoodwinked into acting for them. It has many of the hallmarks of a foreign intel information operation.

Correct, and using heat-seeking missiles close to a burning oil refinery is just plain stupid. bsky.app/profile/spec...

I would really like to learn more about the intersection between data visualization and user experience design. Because if you want data you can explore, then you’ll need both, and right now I’m making stuff up, and really craving inspo pics. Something like Star Trek LCARS would be awesome 😍

"As soon as the Administration bought the warehouses, local communities began to rebel, including in conservative areas..." Yep. And they changed history by doing so. www.nytimes.com/2026/06/18/u...

Whoops 😬 Also I thought that picture of the roof flying was fake at first, look at it, it’s wild.

well done

It’s true

In countries that have successfully reversed democratic backsliding in recent years—from Poland in 2023 to Hungary in 2026—opposition movements made prosecuting those who have broken laws a central campaign issue.

NEW: WIRED investigates how the Thiel-world Dialog society, whose retreats are built around heavy policy discussions with Nobel laureates and military leaders, is secretly fixated on using AI to rank its members based on their wealth and instagram followings.

I’ve seen the paper. It’s not a jailbreak. It was Defense Oriented Prompting (DOP) - a capability defenders need. My thoughts about the hasty Export Controls that made Anthropic halt access to Fable. If national defense is the goal, this is an own goal against us www.wsj.com/tech/ai/anth...

I wrote about what was actually in that #Fable guardrail bypass research paper, and why it should never have triggered an #AI model export control. We can't export control our way to cyber resilience. So many tshirt ideas. www.lutasecurity.com/post/the-fab...

«Anthropic recently shared a third-party research paper on Fable 5 guardrail bypass techniques with me privately and asked for my take.» @k8em0.bsky.social

This made me smile. Maybe if you like music you’ll like it too.

Many fuzzers weren't very good, although some like AFL were definitely game changers. But the main difference is that going from *vulnerability* to a *working exploit* used to be a very expensive art, giving defenders days/weeks/years to react. Now the exploit dev can be near-trivial.

Security researchers, I need some perspective here, I find some of these discussions of Mythos etc difficult to judge because I feel that you could’ve made very similar arguments around fuzzing 10 years ago. This doesn’t feel any more transformative than fuzzing to me, though fuzzing was a big step

“Our findings indicate that LLMs can reinforce or exacerbate existing gender imbalances in scholarly recognition.”

we can solve this