Tutto ciò che fate sul Fediverso è pubblicamente disponibile e archiviabile da Meta, dalle AI, dai Big Tech che possono usarlo in qualsiasi modo.

Meta ha archiviato illegalmente 160TB di libri per adestrare l'AI, scaricandoli online.

Il “silenziare” o "bloccare" istanze non vi salverà mai dai predatori di dati online, serve che tutti facciano in modo di non lasciare troppi dati personali su mastodon.

Cancellare automaticamente i vecchi post può aiutare: https://mastodon.uno/statuses_cleanup

@fediverso

If you run a Nextcloud instance and you also use it for writing notes/to-do lists/tasks, etc. you should really try QOwnNotes.



It's a really good and underrated piece of software, and it even has vim mode.

GIMP 3.0 is released, check it out!

https://www.gimp.org/news/2025/03/16/gimp-3-0-released/

A huge THANK YOU to everyone who contributed in any way - from testing and submitting bug reports through to designing, coding, fixing, packaging, testing some more, translating, documenting, hosting, administration, so many people, so much work, so much to be thankful for!

Welcome to GIMP 3.0!

Hah, the French govt, with some help from Germany, is building an open source Notion clone. Fun!

https://github.com/suitenumerique/docs?tab=readme-ov-file

AI #search engines cite incorrect sources at an alarming 60% rate, study says

Misinformation has always been present on the internet, but #AI certainly isn't helping here.

Additionally, many AI tools ignored Robot Exclusion Protocol settings. AKA they said that robots.txt doesn't apply to them.

(Grok 3's error rate is 94%, lol.)

#misinformation

https://arstechnica.com/ai/2025/03/ai-search-engines-give-incorrect-answers-at-an-alarming-60-rate-study-says/

You may not care about this for stock photos but for any artists out there I encourage you to come up with a way to inject your watermark into your art. I've seen a few people who use there signature as part of how they draw hair, it'll take some effort but it's probably worth it

"The problem isn’t that the fediverse isn’t viable. The problem isn’t that it’s “too complicated.” The problem is that the giants of Silicon Valley have spent 20 years convincing us that anything outside their control isn’t worth our time."

"The internet was meant to belong to everyone. And it still fucking can."

By Daojoan@mastodon.social

https://www.joanwestenberg.com/the-fediverse-isnt-the-future-its-the-present-weve-been-denied/

With GIMP 3 releasing of course the stupid name arguments start up again, if they want to change it they will, if they don't they won't. A name change won't make the application any more popular, it'll still be GIMP and it'll still be in the same state functionality wise. But it's so much easier to argue about a name than it is to address the hard stuff. Regarding the projects stance have a read of the FAQ https://www.gimp.org/docs/userfaq.html#i-dont-like-the-name-gimp-will-you-change-it

This is the last I'm going to say on it, see you in a few years.

The new Fedora KDE Plasma Desktop Edition webpage, I made is online. Currently only the beta is released but the final release is afaik planned mid April.

https://fedoraproject.org/kde/

#kde #fedora

Ever heard of "Federation-Washing"?



https://tube.kockatoo.org/w/kSCtZK8HVhK1xGhf54d6kJ

Microsoft sent me an email telling me to throw my computer in the trash. 🤷

Happy Jellyfin user here! ;) https://www.jeffgeerling.com/blog/2022/why-i-use-jellyfin-my-home-media-library

Vibe coding is just AI copypasta

- Stolen from my Twitch chat

Instagramista on näköjään tulossa pelkkä AI-sovellus. Artikkelissa on kuvakaappauksia siitä, että joillain käyttäjillä näkyy mahdollisuus kirjoittaa kommentteja tekoälyn avulla. (Testiominaisuus vielä siis.)

Saahan AI-kommenteilla somesta vielä vähän nopeampaa, kun ei tarvitse pysähtyä ajattelemaan, että mitä ihmisenä kirjoittaisi toiselle ihmiselle. Tehokkuus ja sisällöntuottaminen on toki tärkeämpää kuin aitojen kontaktien ja keskustelujen mahdollistaminen käyttäjille... 🙄

No Metahan on puhunut tästä omasta mahtavasta tekoälyprojektistaan jo jonkin aikaa, että ei sinänsä pitäisi yllättää, että sitä halutaan tunkea joka paikkaan.

Tulipa myös mieleen seuraavaa:

Jos henkilö A käyttää tekoälyä kommentin keksimisen henkilön B kuvasta.

Artikkelin kuvakappauksessa luki, että käyttämällä työkalua hyväksyy tekoälyehdot ja että tiedot "saattaa" siirtyä Meta AI:n hyödynnettäväksi. Jos henkilö A on siis kieltänyt aiemmin sen, että Meta ei voi käyttää A:n sisältöä tekoälyn kouluttamiseen, niin voikohan tuossa tapauksessa käydä niin, että AI-kommentteja tekemällä sen henkilön A kieltokin jotenkin purkautuu, koska "hyväksyy ne ehdot"? En yllättyisi, jos näin olisi, koska Meta haluaa jengiltä kaiken datan.

#sosiaalinenmedia #instagram #tekoäly #meta

https://www.socialmediatoday.com/news/instagram-tests-ai-generated-comments-posts/742633/

#eBay now trains internal and third-party #AI on user’s personal data

https://discuss.techlore.tech/t/ebay-now-trains-internal-and-third-party-ai-on-users-personal-data/12993

#privacy #ecommerce #shopping

Open Source Infrastructure has an AI problem



https://tube.kockatoo.org/w/woce36hTzF6JdnBiBEaRZJ

And nice summary from niccolove@fosstodon.org about the current situation https://thelibre.news/foss-infrastructure-is-under-attack-by-ai-companies/

AI scraping bots are actively damaging the web. They don’t follow rules, they’re careless and sloppy, and they pretend to be regular web browsers. Genuine users and bots are being hurt by this, because it’s a DDoS attack that knows how to fly just below the radar at all times.

This is an unprecedented attack on the open web. It’s not just a few kids with some hacked servers - it’s startups motivated by greed.

You can’t tell them to knock it off, because they’re anonymous cowards. You can file an abuse report with their host, but they’ll just go set up elsewhere.

https://youtu.be/cQk2mPcAAWo

We've entered the Era of the AI wars, AI scrapers are openly ignoring robots.txt and other measures designed to tell bot website visitors to behave in a certain way.

If that's how it's going to be let's get into an arms race then. There's open source tools like Anubis, Cloudflare announced AI Labyrinth. This won't be the end of AI scrapers, they will get better and so will these tools and so many resources will be wasted on a problem that was already resolved if the bots worked properly

How Meta and OpenAI pirated the world's books to train their LLMs on high quality written text in multiple languages - because legal access, they felt, would be too slow and costly.

https://www.theatlantic.com/technology/archive/2025/03/libgen-meta-openai/682093/