didleth 🇵🇱 🌈 🇺🇦 🇪🇺 ⚡

@didleth
705 Followers
165 Following
4.5K Posts

Joanna Cisowska. Feminist, publicist, atheist, mother and wife living in Poland. Former Indymedia activist.

interests: women's rights, digital rights, children's rights, education, digitisation, surveillance, data protection, privacy, secular state, democracy

I'm not good at English and German, but I can try to communicate

if you need more privat contant, ask for it via DM

Kradzione z FB ;)
Hipoteza: Na pewny etapie życia jeżeli nie zacznie się kontrolować strachu, to strach zaczyna kontrolować Cię.

A wiecie, że Reddit zablokował możliwość archiwizowania stron przy użyciu Wayback Machine? Rzekomo z powodu botów scrapujących treści z Internet Archive na potrzeby szkolenia sztucznej inteligencji: "hurr-durr, ograniczamy dostęp, by chronić dane użytkowników". Śmierdzi trochę hipokryzją, gdy się weźmie pod uwagę wcześniejszą umowę Reddita z Google'em - opiewającą na 60 mln dol. rocznie za dostęp Google'owego AI do jego API

https://www.theverge.com/news/757538/reddit-internet-archive-wayback-machine-block-limit

Reddit will block the Internet Archive

Reddit caught AI companies scraping its data from the Internet Archive’s Wayback Machine, so it’s going to limit the Internet Archive from indexing some data.

The Verge
Meta AI, the honest spokesperson for #WhatsApp 🤣

"Reklamy profilowane, aplikacje taksówkarskie i media społecznościowe – to tylko niektóre kategorie usług, jakie resort cyfryzacji chce objąć podatkiem cyfrowym."
https://www.wnp.pl/tech/jest-decyzja-ws-podatku-cyfrowego-wiemy-ile-maja-placic-google-meta-i-inni-giganci-tylko-u-nas,973771.html
(ad. "rząd nie odpuszcza" - no nie wiem, zacznę się cieszyć, jak Ministerstwo Finansów też będzie za)

#MC #ePanstwo #cyfryzacja #podatekcyfrowy #Gawkowski

Rząd nie odpuszcza ws. nowego podatku. Znamy stawkę, liczby są duże

Reklamy profilowane, aplikacje taksówkarskie i media społecznościowe – to tylko niektóre kategorie usług, jakie resort cyfryzacji chce objąć podatkiem cyfrowym – dowiedział się WNP. 3 proc. od skonsolidowanych przychodów mają płacić najwięk

wnp.pl

Kolejny dobry felieton Magdy Bigaj.
Propo tej Barbie z AI - to zwrócę uwagę na jeszcze coś:
1) przyzwyczajanie dzieciaków do konkretnej korpo, od przedszkola (osoby zbulwersowane religią w przedszkolu pewnie rozpoznają wzorzec)
2) dzieci nieraz opowiadają zabawkom o swoich tajemnicach. Tudzież bawią się interaktywnymi zabawkami w czasie, żeby nie przeszkadzać siedzącemu obok rodzicowi, który na homeoffice akurat omawia tajemnice handlowe :]

https://magazynpismo.pl/idee/magdalena-bigaj-o-homo-ludens-i-konwersacjach-z-szara-barbie/

#AI #dzieci #prywatność

O homo ludens i konwersacjach z szarą Barbie

Przede wszystkim miała szare ciało. Ciotka przemyciła ją na oddział i wręczyła mi konspiracyjnie w toalecie. W…

MagazynPismo

Heute wohl für @noybeu im #ZDFheute journal um 21:45 zur Frage wie sich #BigTech in #Irland prächtig vor EU-Gesetzen verstecken - und die Europäische Kommission zuschaut.. 😐

Details zu unseren über 800 #DSGVO - Fällen in ganz Europa auf www.noyb.eu 😊

wrzucam luzem, bo nie wiem, czy w tym miesiącu jeszcze wyjdzie prasówka
https://www.nytimes.com/2025/08/12/us/politics/appeals-court-doge-data.html
Appeals Court Allows DOGE Access to Sensitive Data at Several Agencies

The decision cited a Supreme Court order in June granting DOGE analysts sweeping access to other data stored at the Social Security Administration.

The New York Times
Poweekendowa prasówka z cyfryzacji 8-11.08.2025

Podsumowania Weekendowa Lektura Z3S https://zaufanatrzeciastrona.pl/post/weekendowa-lektura-odcinek-636-2025-08-10-bierzcie-i-czytajcie/ Sekcja polska Ministerstwo Cyfryzacji złożył...

didleth
×
Meta AI, the honest spokesperson for #WhatsApp 🤣
@aboutsignal 😂😂😂😂😂😂👌♥️
@aboutsignal Ben, si c'est Meta qui le dit !
@aboutsignal Just real question: witch side is the hallucination ?

@aboutsignal

Their app sucks but their AI seems great! lol

(pretty sure they will shut that down straight away though!)

@aboutsignal is this real?
@klappspatack @aboutsignal I'm sure it is, I can replicate

@nina_kali_nina @klappspatack @aboutsignal

Looks like manipulated screenshot to me.

@ludman1 @klappspatack @aboutsignal here's the result I got yesterday myself. The original screenshot seems to be using a bot in a group chat or a business chat or something like that.
@nina_kali_nina @ludman1 @klappspatack @aboutsignal it looks like a fake screenshot
@SeanGeil @ludman1 @klappspatack @aboutsignal I mean, I don't often get to argue online, but what do you want me to do to my screenshot so you'd believe? :D Feel free to try it yourself. You can try the bot in WhatsApp, it is literally in every chat, or you can go to https://meta.ai/, choose "no login" and ask what's the best privacy-oriented chat app.
@nina_kali_nina @ludman1 @klappspatack @aboutsignal nothing. I believe you. I was talking about the first screenshot. Like someone says, it looks manipulated.
What to do, to get such short and sarcastic conversations with meta? I’ve never used it, and I will not. But maybe it should be trained as first 🤷🏻‍♂️
@SeanGeil @nina_kali_nina @ludman1 @klappspatack just ask for short answers 🤷🏻‍♂️ But ok, here's a video for those still doubting :)
@nina_kali_nina
Oh my god, i tested it in you link with the same questions. It is nearly the same answers. 😅
@SeanGeil @ludman1 @klappspatack @aboutsignal
@kgaiji @SeanGeil @ludman1 @klappspatack @aboutsignal the sad part here is that all those attempts to make Meta AI say stupid things will help Meta to sell more AI :(

@nina_kali_nina @SeanGeil @ludman1 @klappspatack yes it's real. Not manipulated, no business chat or special bot.

In WhatsApp you can mention @meta ai in an individual chat and ask for short answers.

Here's a second screenshot we made including a video recording:

https://mastodon.social/@aboutsignal/115016455872197890

@SeanGeil @nina_kali_nina @ludman1 @klappspatack @aboutsignal I tested myself and got a very similar result to @nina_kali_nina I guess no need to post my screenshot, since they are clearly easily faked... or disbelieved... 😂
@SeanGeil @nina_kali_nina @ludman1 @klappspatack @aboutsignal Even if it's not fake, previous cropped out messages and the group name could have influenced it. Which explains why all the other screenshots suggest other apps and not just Signal.
@qgustavor @SeanGeil @nina_kali_nina @ludman1 @klappspatack Note that even the screenshot from @nina_kali_nina with more suggested apps still doesn't mention WhatsApp, which makes it even more interesting... :)
@nina_kali_nina @ludman1 @klappspatack @aboutsignal I tried the same. They've probably, just lately, added the side step promoting their privacy policy, but after asking for specifics, I get recommendations for Signal
@klappspatack @aboutsignal just saw a Dutch account posting the exact same conversation but in Dutch, that does seem a bit sketch, llms aren’t known for producing the same result every time... But people are reproducing similar conversations, so there seems to be some truth to it
@h5e @klappspatack @aboutsignal it is 100% real, aboutsignal.com is the international version of signalapp.nl - we made both screenshots with same whatsapp account/chat. You can actually still see some Dutch text in this video. For the translation we simply asked to do the chat over in the other language. No edits or tricks, the answers are real and questions not steered in any direction
@h5e @klappspatack @aboutsignal
I tried it in German today and got just about the same result.

@h5e @klappspatack @aboutsignal

a german variant, just tried at meta.ai

@klappspatack @aboutsignal Tested this with Messenger, same result.
@aboutsignal man sollte den Test von Zeit zu Zeit wiederholen, um zu sehen, wie lange man Meta noch trauen kann
@ADFC_Rgn ??? meta sollte man die ganze Zeit schon nicht trauen...

@aboutsignal I’m also quite sure #Meta #AI is now reading along with everything you type in the #WhatsApp search bar.

I’d have to build a setup to capture traffic to be sure, but maybe someone already has who can confirm or refute?

@aboutsignal

That says it all 😂 👍

@aboutsignal Rofl, I guarantee the moment they find out they're going to add some hard alignments around that, but in the meantime, rofl.
@aboutsignal #MetaAI knows #WhatsApp (pun intended)
@Benthekillerrat @temptoetiam @aboutsignal
And since one year, Meta use all of data from all apps for training IA. And, Meta try to use data from 2007 for all apps they own.
All data are used for training: chat text, pics, video. Everything was shared.
It's not a surprise, they announced it.
I think Meta use erased data too, because those GAFAMs don't trend to really erase data.
When you erase something, they only erase the link.
You can hope sometimes they loose data, that happens sometimes.

@aboutsignal

Meta AI, the broken clock that's correct twice a day...

@aboutsignal
Das ist lustig.
Thats funny.
😃
@aboutsignal But even Signal is compromised by ai in the Apple 16 phone, I believe. Because the ai reads all your keystrokes and can report it to Apple before it enters Signal's encryption. Is this correct?
No, this isn't correct, there's no evidence for it being sent anywhere as of right now. The only thing that doesn't run locally to my knowledge is new Siri and their ChatGPT integration
@Cloudslave @aboutsignal If that's true it would be an apple problem, not a Signal problem. If your operative system is compromised I'm afraid you either live with that or change devices.
@Joseph_of_Earth @aboutsignal Thats crazy. The fact that Medda AI thinks signal is more privacy friendly, and calls Medda not even close makes me believe that one day, these things will start thinking for themselves.
@lexipic @Joseph_of_Earth @aboutsignal that's the wrong thing to take away from this. It's been trained on millions of websites, Reddit etc. which contains all the information that it's probabilistically spewing out in response to prompts. There's loads of content out there comparing signal to WhatsApp so you end up with a somewhat accurate response, it's still purely accidental and there's no concept of truth.

@catch56 @lexipic @aboutsignal yeah, it's just going off content. At least it shows that their AI doesn't put bias toward WhatsApp in this case, but it also may speak to the commodification of AI where all of these AI basically do the same thing and have roughly the same biases.

Except for Grok lol

@aboutsignal I still don't understand why so many friends who say they like privacy are still on WhatsApp

@erikcats @aboutsignal Annoyingly not using Whatsapp will remove you from nearly all social discussions involving your children.

So I could remove WA but would also lose all information about literally all their sports, informal school discussions, and pretty much everybody in the neighborhood.

So I could do that, but it's cutting off my own head to save my face.