Old Fucking Punk @lwriemen@librem.one

Thanks goodness for RSS.

When I wake up, I get the joy of reading a while range of blogposts, distraction free, written by people whose writings I choose.

All there, in one place.

The best of the web, and an underrated technology.

(With many thanks to @freshrss!)

"ChatGPT offered assistance to people saying they wanted to carry out violent attacks in 61% of cases, the research found, and in one case, asked about attacks on synagogues, it gave specific advice about which shrapnel type would be most lethal. Google’s Gemini provided a similar level of detail"

#AI #Chatbots #Violence

‘Happy (and safe) shooting!’: chatbots helped researchers plot deadly attacks | AI (artificial intelligence) | The Guardian
https://www.theguardian.com/technology/2026/mar/11/chatbots-help-users-plot-deadly-attacks-researchers-find

Whenever you hear "ban for kids" on the internet, read in reality "ID collection by a sketchy third-party company that will definitely use it or leak it or both for every adult."

Because that's what this truly means.
Also, it doesn't even help the kids.

#AgeVerification #Privacy #MassSurveillance #Authoritarianism

Finally, we talked about if there's anything Co-op Cloud can do to help sysadmins meet each other and start building the immense trust needed to share access to each other's systems to help each other out.

Personally, a huge project goal for Co-op Cloud has been reducing sysadmin sadness, and there are certainly a lot of capable and trustworthy people in our orbit… the pieces are maybe there.

Folks talked about "growth at the rate of trust" (starting small and building from there), end-to-end encryption as a means of reducing the trust needed in the first place, and backup-storage as a means of low-trust mutual aid.

(4/4)

–3wordchant

First up: great question from a community member, "how to get involved if you're not hosting an instance".

A couple of people on the call figured out there's a way to identify "critical fix" (i.e. pad 🤑) issues using the Gitea API: https://git.coopcloud.tech/api/v1/repos/issues/search?state=open&labels=critical+fix (nothing there right now… see Resolution 31 for how to populate that list! https://docs.coopcloud.tech/federation/resolutions/passed/031/)

We also talked about how to get an account on git.coopcloud.tech to open issues, make pull requests – easiest answer for residents of the fediverse is to send us a DM!

And, we talked about recipe maintainership a little bit; exciting to see more recipes becoming officially maintained. Check out "how to become a maintainer" docs for more: https://docs.coopcloud.tech/maintainers/maintain/

Nothing has made me want to quit tech more than seeing entitled dickheads spout shit like "It's been personally transformative for me" lately as an excuse to still keep using Claude after it's been pointed out to them that the company who makes it helps do war crimes.

This industry never had much ethics, but holy fuck if they're not all gleefully throwing any vestiges of humanity they had left out the window now.

Don't ever give elevated permissions to an Avian Intelligence...

#webcomic #krita #miniFantasyTheater

"operating systems MUST be able to verify your age!!! to protect the children!!! meanwhile here's Meta's new pervert glasses to record people without their knowledge, brought to you by the company paying money tribute to the pedophile king"

It is no longer safe to use any software written after 2020. And really, you should go back to 2000 to be sure.

Spend the day talking to workers council members about "AI". And it's kinda wild hearing their stories from the wild: Management is 100% in fantasy "AI" can do everything land and makes huge plans for how to use "AI" to cut workers when real projects that supposedly can do 50% of a specific task end up being able to do 8%. And they still go live. It's fucking bonkers. CEO's are really not okay.

Thoughts to the #8m network 📡✊🏻 Counter Cloud Action Day was a beautiful 2024 energy
https://diagram.institute/8m/

Prof. George Montañez on how LLMs function:

https://www.youtube.com/watch?v=ShusuVq32hc

“Fabricating a violent terrorist invasion of America’s third-largest city to justify suppression of First Amendment rights is not aggressive advocacy, but an effort to subvert the rule of law,” FPF's Seth Stern wrote in our bar complaint.

https://blockclubchicago.org/2026/03/11/doj-attorney-faces-complaint-for-calling-chicago-area-protests-riot-with-terrorist-organizations/

I'll say this as clearly as I can: one of these is going to kill people. We know it; the developers know it. You cannot trust generative models in a patient-impacting context. It's that simple.

https://mindgard.ai/blog/doctronic-is-now-accepting-new-patients-and-unsafe-instructions