Discussion
URGENT: Autistici/Inventati says Banca Etica is terminating its banking relationship and blocking access to donated funds, leaving A/I unable to pay suppliers or maintain services. A/I says it will pursue legal action.
Read. Share. Preserve. Organize.
Read A/I’s Sept. 5 statement
https://keepitfree.ai/announcements/banca-etica-shuts-down-a/i-funds/
San Sebastian seems nice 🥲🌴💸 #SanSebastian #Donostia
@Gina Sandy beach bordering on high-rise apartment/hotel blocks is nice? That beach looks like a fleet of maintenance tractors come every morning sifting the trash and worse out of the sand.
@alice_pea_3526 I wish. they still have their bitcoin miner and NFT shilling from 2021-2022 hanging out.
@davidgerard @alice_pea_3526 Let's Encrypt is getting people LLM scraped.
Bitcoin Core had a chance to embrace capabilities-based security, the foundation of a strong anti-"agent" immune system. It didn't.
Because some technical folks* like to complain about #XMPP...
Every single problem in XMPP can happen in any other decentralized protocol as well. Hype can die, funds can dry up, implementations can stagnate, walled gardens can restrict what the implementations can do, design decisions can turn out to be the wrong ones years down the line (necessitating possibly unaesthetic hacks)...
At some point we all have to put our foot down and support established standards, no matter how (temporarily) inconvenient. Users, testers, and developers must do everything they can to help.
*The non-geeks are somehow never affected by the geek issues...or they don't really care ¯_(ツ)_/¯
Training AI models on copyrighted works is "squarely what transformative use is," EFF’s Tori Noble told Reason - these are "general purpose tools" and "exactly the kind of creation that copyright is designed to foster.” https://reason.com/2026/09/04/doj-says-barring-ai-training-on-copyrighted-material-could-severely-hamper-tech-progress/
@eff I'd be interested to know what you, @pluralistic, think about it?
OpenAI commits $1B to AI cyber defense for critical infrastructure
OpenAI has committed $1 billion in subsidized access to its AI cybersecurity tools, training and technical support through its new Daybreak for Frontline Defenders initiative, targeting organizations that protect critical infrastructure and essential services....
https://itnerd.blog/2026/09/05/openai-commits-1b-to-ai-cyber-defense-for-critical-infrastructure/
Just a thought: algorithmic social media is doing the same thing to online communities that cars did to cities. It is creating little pockets of isolated individuals in a sea of people instead of creating communities. It gives people an impression of being surrounded by people without actually forming real connections
To celebrate that TutaCrypt has been formally verified by University of Wuppertal we are having a Tuta hoodie giveaway 🎉
To stand a chance to win, let us why you chose Tuta to secure your emails in the comments below 👇
P.S: We hope you are as happy about our hoodies as we are 🥰
@Tutanota open-source and secure, what not to like.
the main email providers you can be sure are scanning your emails in LLM servers.
Shared infrastructure is how the #Fediverse can grow without recreating enshittified platform silos.
We’re excited to share that Bonfire is working with Mastodon on two critical areas of Fediverse infrastructure: remote media storage and automated content detection.
Built around open Fediverse Auxiliary Service Provider (FASP) specifications, this work will help platforms connect to shared services instead of each having to build and operate the same capabilities alone.
Our aim is to make these capabilities open, interoperable, and available for any federated server to adopt.
A big thank you to @Mastodon@mastodon.social team for involving us in this work. We’re looking forward to building together and sharing what we learn along the way.
Read the full post: bonfirenetworks.org/posts/bu...
#ActivityPub #FASP
A whole season of #TedLasso being dedicated to this issue, btw.
From: @researchbuzz
https://researchbuzz.masto.host/@researchbuzz/117218262818989477
Je viens de me décider à me remettre à RSS ! C'est quoi votre top 3 ou top 5 de fils RSS ? (à part standblog.org ;-)
@nitot le mien c'est emploi.afjv.com 😂 . En vrai je m'y suis remis mais via l'app glance que j'adore !
https://github.com/glanceapp/glance
I’m honestly curious, how should an author earn money while making writing their primary economic activity? Or writing shouldn’t be an economic activity at all?
So hang on. The screenshot test says I can pay 10 cents for a robot to move my mouse to a button?
That can't be right, it must be 10 cents to run the whole benchmark, there are only 1500 samples per run.
But the output tokens are all less than 1000... So either it can move my mouse with fewer than 1 token on average... Or...... 10 cents to move my mouse is really what is being shown here.............
@jonny but, hear me out, for that 10 cents each 10th time robot will move your mouse wrongly!
Today, I will be mostly playing hide and seek in the woods.
Officially, it is Airsoft, but it is going to be a rather different game, I think.
A small number of us (5-10, I am guessing), on a large site, probably playing the role of very small units behind enemy lines, with objectives to carry out.
But with a break for lunch.
15 of us. In 35ish acres.
And I didn't sit on my pasty.
Holy God this is _perfect_.
This isn't a word!!! Not in any English dictionary I've ever seen 🙂↔️
@jaz
@gomez
@siaronj
@philroyle
@TCMuffin
@lexicaleigh
@morchwyn
@GnomeBones
@alexisbushnell
@epistatacadam
#Fedle #665 3/6
🟩🟩⬛⬛🟩⬛
🟩🟩🟩⬛🟩🟩
🟩🟩🟩🟩🟩🟩
Sorry, the app encountered an unexpected error
Install bonfire.cafe
Get the full app experience
![[gpt-6 achieving 90% accuracy between 7 and 11 cents, gpt-5.5 60-80% accuracy between 3 and 4 cents]
ScreenSpot-Pro tests whether models can locate the correct interface element in high-resolution screenshots of professional software. GPT‑6 Astra sets a new high with 92.7% accuracy.](https://media.neuromatch.social/media_attachments/files/117/217/770/627/753/648/original/9ac661ab980533e2.jpg)
![[gpt-6 at 90% accuracy between 0 and 750 output tokens. gpt-5.6 at 60-80% between 100 and 1000 output tokens]](https://media.neuromatch.social/media_attachments/files/117/217/770/663/369/696/original/7c2a69a50c869032.jpg)
