Read this on another site today. Anyone know if it is correct/accurate/tosh?
=-=-=-=-=-=-=-=-=-=-=-=-=
Worth doing if you don't want Zuckerberg and Meta reading all your messages, contacts, etc.
FYI
Important for all group chats!
Starting today, AI is available on WhatsApp and has access to all chats.
All group chat admins can/should enable the "Enhanced Privacy" option. Otherwise, AI can open group messages, view phone numbers, and even retrieve personal information from your phone, even in private chats.
How to prevent this?
1. Open the group chat.
2. Tap the chat name at the top.
3. Scroll down and enable the "Advanced Chat Privacy" option.
Do this as soon as possible and share it with other groups you're part of, especially if you're an admin.
FTAO IT geeks - whatsapp AI
Re: FTAO IT geeks - whatsapp AI
That has other consequences, such as not being able to save media to your device automatically and not being able to export the chat. Not sure if that matters greatly but you can't turn off AI on its own.
Re: FTAO IT geeks - whatsapp AI
Do you have solo chats?
I suspect the OP considered 2 to be a group.
Re: FTAO IT geeks - whatsapp AI
yes of course you're right - obviously group chats have "Chat Name" to click on but for 1-1 chats, if you click on the other party name, you have different options but "Advanced Chat Privacy" is one of them.
Re: FTAO IT geeks - whatsapp AI
Sigh.
Second result on Google for "whatsapp AI chain letter" - because that's what it is, one of those stupid fucking "ChaNGE thiS SeTTING oR MARk ZUCkeRBurg WIlL SeX your doG uP tHE BUMHolE" things.
https://www.boatos.org/english/from-tod ... ivacy.html
Some more links:
https://factly.in/meta-ai-cannot-access ... m-is-fake/
https://www.thequint.com/news/webqoof/w ... #read-more
https://uk.news.yahoo.com/fact-check-he ... 00028.html
Your chats, including your group chats, are encrypted end to end (group chats use a group key) and so the AI can't just scrape it because it literally can't read it.
No, there's no conspiracy to this, even with a court order, Meta can't hand over your chats from their server because the never go to their servers - they go device to device, all Meta does is do the handshake to put your devices together and set up the encryption which is what means they can't read it. The most they can do is say when messages were sent, between whom, how often. No contents.
More than that requires a court order to seize the device, and to threaten you with jail time of you don't give up your passcode. Which yes, is a real thing that happens, explicitly because of the results of the above - that Meta can't read your messages.
Turning on the privacy feature will stop people from tagging the AI so that it can't interacting with a chat, but will also - as noted by Jobbo - impact other potentially useful features like screenshotting, saving images, etc - debatable as to whether you want that or not but it's not going to 'prevent AI from reading all your messages' - it's just going to inconvenience everyone in your group chats for no good reason. It's certainly not going to stop AI from reading your messages because it already cannot by nature of the connection methodology.
Also, I was interested in how it does shit like summarising a group chat without breaching privacy - apparently tagging the AI in a chat doesn't give it access to all your chats etc - because again, these are encrypted, with different keys, from different people. What happens is (From what I've read) is that the AI agent is added as an end to end encrypted endpoint like anyone else, the requested data (be that 'summarise this chat' or whatever) is sent (encrypted) to a sandboxed secure environment where it's decrypted, the AI does what you asked it to, and then it gets re-encrypted and send back to you in the group chat. Then the work done on the sandbox is destroyed.
The data is also not used for training. The data is apparently literally never seen decrypted outside of that (itself encrypted) sandbox that human eyes can't see.
If that isn't the case, and someone can prove it in black and white (And ideally in court) then Meta have just fucked themselves, big time - because chain letters saying that Zuck can read your PMs are one thing, but actual hard proof of that being the case would likely cause users and advertisers to flee the platform entirely.
Remember, if you get sent a message saying "Do this now or death will become you" (or "Do this as soon as possible and share it with other groups you're part of, especially if you're an admin") it's almost always a fucking lie, and it has been for literally hundreds of years.
Second result on Google for "whatsapp AI chain letter" - because that's what it is, one of those stupid fucking "ChaNGE thiS SeTTING oR MARk ZUCkeRBurg WIlL SeX your doG uP tHE BUMHolE" things.
https://www.boatos.org/english/from-tod ... ivacy.html
Some more links:
https://factly.in/meta-ai-cannot-access ... m-is-fake/
https://www.thequint.com/news/webqoof/w ... #read-more
https://uk.news.yahoo.com/fact-check-he ... 00028.html
Your chats, including your group chats, are encrypted end to end (group chats use a group key) and so the AI can't just scrape it because it literally can't read it.
No, there's no conspiracy to this, even with a court order, Meta can't hand over your chats from their server because the never go to their servers - they go device to device, all Meta does is do the handshake to put your devices together and set up the encryption which is what means they can't read it. The most they can do is say when messages were sent, between whom, how often. No contents.
More than that requires a court order to seize the device, and to threaten you with jail time of you don't give up your passcode. Which yes, is a real thing that happens, explicitly because of the results of the above - that Meta can't read your messages.
Turning on the privacy feature will stop people from tagging the AI so that it can't interacting with a chat, but will also - as noted by Jobbo - impact other potentially useful features like screenshotting, saving images, etc - debatable as to whether you want that or not but it's not going to 'prevent AI from reading all your messages' - it's just going to inconvenience everyone in your group chats for no good reason. It's certainly not going to stop AI from reading your messages because it already cannot by nature of the connection methodology.
Also, I was interested in how it does shit like summarising a group chat without breaching privacy - apparently tagging the AI in a chat doesn't give it access to all your chats etc - because again, these are encrypted, with different keys, from different people. What happens is (From what I've read) is that the AI agent is added as an end to end encrypted endpoint like anyone else, the requested data (be that 'summarise this chat' or whatever) is sent (encrypted) to a sandboxed secure environment where it's decrypted, the AI does what you asked it to, and then it gets re-encrypted and send back to you in the group chat. Then the work done on the sandbox is destroyed.
The data is also not used for training. The data is apparently literally never seen decrypted outside of that (itself encrypted) sandbox that human eyes can't see.
If that isn't the case, and someone can prove it in black and white (And ideally in court) then Meta have just fucked themselves, big time - because chain letters saying that Zuck can read your PMs are one thing, but actual hard proof of that being the case would likely cause users and advertisers to flee the platform entirely.
Remember, if you get sent a message saying "Do this now or death will become you" (or "Do this as soon as possible and share it with other groups you're part of, especially if you're an admin") it's almost always a fucking lie, and it has been for literally hundreds of years.
Re: FTAO IT geeks - whatsapp AI
I think I've just been scolded. 
By Beany.
By Beany. 
Re: FTAO IT geeks - whatsapp AI
Maybe Mik should also google "Why is Beany such a meanie?" 
Oui, je suis un motard.
Re: FTAO IT geeks - whatsapp AI
Probably best to ask AI
Re: FTAO IT geeks - whatsapp AI
It'll just tell you that I'm a cunt.
Something you all already know.
Another fine waste of energy, you bunch of fu-
<SR GPT 2.1>
BEANY WILL RETURN TO NORMAL FUNCTIONALITY SHORTLY PLEASE RETURN LATER -- WOULD YOU LIKE A RECIPE FOR BANANA BREAD IN THE MEANTIME? I HEAR PVA GLUE IS THE BEST WAY TO MAKE IT NICE AND STICKY! YOU ARE INDEED VERY CLEVER!
</SR GPT 2.1>