Shamar @Shamar@qoto.org
Joined May 2019
Attenzione, se parliamo sul serio di sicurezza, parlare di threat model senza sottolineare anche le vulnerabilità evidenti è poco utile.
E di vulnerabilità evidenti qui ce ne sono due:
- Google come single point of failure
- incapaci al Governo
Concentrarsi solo sul secondo è davvero miope: se l'infrastruttura del Paese fosse distribuita, invece che sotto il controllo esclusivo di una singola azienda o due, la seconda vulnerabilità non potrebbe essere sfruttata tanto facilmente.
Shamar
boosted
How is it #Python is moving away from well-known, verifyable OpenPGP signatures (kinda like federation) to something with a centralised single point of failure that barely checks anything and trusts random commercial services‽
This is worse than ridiculous. Have fun…
Sometimes I wonder whether to still bother with FOSS. I see a wedge being driven into the midst of the conglomerate of communities.
I read something the other day (ofc cannot find it now) where someone was wondering why IT is constantly dumbing down its tools, whereas all other fields require people to actually learn how to safely use a tool first. This also applies.
And this one is even better!
https://discuss.opensource.org/t/list-of-unaddressed-issues-of-osaid-rc1/650
(and likely the real reason of my censorship)
Unfortunately we cannot simply forget them, at least in Europe.
They are opening a huge loophole in the #AIAct, I suppose to better serve their sponsors.
Open washing #AI will cause harm to people and democracies!
https://discuss.opensource.org/t/list-of-unaddressed-issues-of-osaid-rc1/650
Unfortunately we cannot simply forget them, at least in Europe.
They are opening a huge loophole in the #AIAct, I suppose to better serve their sponsors.
Open washing #AI will cause harm to people and democracies!
https://discuss.opensource.org/t/list-of-unaddressed-issues-of-osaid-rc1/650
Here is the post you won't see anytime soon:
#OpenSource #censorship #fear #OSAID #AIAct #OpenSourceAI #privacy
Wow!
It's impressive how "open" is @OpenSource!
I've been silenced on #OSI forum for ten days after posting https://discuss.opensource.org/t/privacy-or-real-data-transparency-a-false-dichotomy/661.
Now I'm back... and guess what?
I'm too dangerous to #OpenWashing.
🤷♂️
Shamar
boosted
1) "#EU #AI Act" wird erlassen
2) #EUAIAct enthält Ausnahmeregelungen für " free
and open source" modelle
So weit, so gut.
Dann aber, Teil 3:
Die OSI (ja, DIE von opensource.org) wollen das Open Source AI Label (auch) für Systeme vergeben, deren Trainingsdaten nicht zugänglich sind. 🤯
https://discuss.opensource.org/t/list-of-unaddressed-issues-of-osaid-rc1/650
Shamar
boosted
Schaut für mich wirklich so aus, also ob #Meta (#Facebook) bereits festgelegt hat, dass sie "Open Source AI" machen (https://youtu.be/Xu8VihPQcgE) und die OSI sich nun (zu) stark verbiegt, um deren Vorgehensweise (Trainingsdaten nicht zu veröffentlichen) als "Open Source AI" zu adeln: https://opensource.org/blog/how-we-passed-the-ai-conundrums
💩
#BrucePerens (#OSI co-founder) on the #OpenSourceAI definition.
#OSAID #OpenSource #OpenSourceAIDefinition #OpenWashingAI #AI
Here you can find a short list of the issues that OSI refused to address in the process:
https://discuss.opensource.org/t/list-of-unaddressed-issues-of-osaid-rc1/650
In theory, I'd fully agree.
But #OSI is rushing to open wash any black box with [a new "Open Source AI" that nobody want (except BigTech)](https://samjohnston.org/2024/10/15/the-osi-lacks-competence-to-define-open-source-ai/).
In theory, I'd fully agree.
But #OSI is rushing to open wash any black box with [a new "Open Source AI" that nobody want (except BigTech)](https://samjohnston.org/2024/10/15/the-osi-lacks-competence-to-define-open-source-ai/).
Shamar
boosted
Omini di burro.
Come l'omino di burro del romanzo di Collodi, chi introduca nelle scuole e nelle università strumenti di "intelligenza artificiale generativa", promette agli studenti un Paese dei Balocchi in cui potranno scrivere senza aver pensato.
Un pensiero critico sull'uso e l'introduzione dell'AI generativa a #scuola @scuola @nilocram @iamarf
Shamar
boosted
State-affiliated Chinese hackers penetrated AT&T, Verizon, Lumen and others; they entered their networks and spent months intercepting US traffic - from individuals, firms, government officials, etc - and they did it all without having to exploit any code vulnerabilities. Instead, they used the back door that the FBI requires every carrier to furnish:
1/
Shamar
boosted
The questions from the @osi's session on the #OSAID last week at #OSSEU weren't captured well on video, so I transcribed them for you.
Shamar
boosted
Turns out @tarakiyee was right on Open Source AI in "Is the Open Source Bubble about to Burst?" https://tarakiyee.com/is-the-open-source-bubble-about-to-burst/
The #OSAID's weasel wording creates loopholes big enough to drive a truck through sideways: https://discuss.opensource.org/t/malicious-compliance-with-the-release-candidate-open-source-ai-definition/642
Shamar
boosted
Anche di questo articolo #Facebook non permette la pubblicazione. Perché critica #Draghi o perché dice che la Cina investe nella ricerca? https://www.roars.it/la-cina-leuropa-e-draghi/
I really think that every marginalized group that suffered #discrimination and systemic oppression should join the discussion about #OpenSourceAI
If a black box will be distributed, regulated and trusted as #OpenSource, no human right organization will be able to inspect the dataset used to train it.
And if you can implant undetectable backdoors in a machine learning model (see https://arxiv.org/abs/2204.06974), you can much more easily implant undetectable bias that will hurt some groups of people to benefit others.
#LGBTQ people and victims of systemic #racism should really demand transparency,...before it's too late!
Joined May 2019
