News for hackers. Stuff that matters.
Re:Meta won't train AI on Euro posts after all, as watchdogs... by ehnto on Monday June 17, 2024 @03:52AM (#40703254)
Privacy is another, if they are training with chat messages in a sense it "knows" the information in your chat. Data exfiltration would be really interesting in this space, a weird blend of tech and social engineering.

I don't really care about the money but as a society we should care a lot about such a blatant heist of personal effort. For many their words online are their works of the day, it's not at all fair for a company to take that and use it, and even more insidious that it will be used to replace the people who created those works.

Re:Meta won't train AI on Euro posts after all, as watchdogs... by stavros on Monday June 17, 2024 @04:02AM (#40703320)
But this is about FB posts, not personal writings, for example. Meta owns your FB posts, don't they?
Re:Meta won't train AI on Euro posts after all, as watchdogs... by ehnto on Wednesday June 19, 2024 @05:29AM (#40726503)
I don't think it is fair to assume your average person understands that their personal rants are ending up in an LLM/Publically accesible AI, even if the T&Cs say they can.

That's what the fuss is about I guess, we are working out the ethics our societies will accept on a pioneering front of technology. I don't think Facebook gets away on the letter of the law this time, they will need to adjust to new regulations as we decide on what's ethical.

Re:Meta won't train AI on Euro posts after all, as watchdogs... by prasoonds on Monday June 17, 2024 @04:03AM (#40703324)
The article very clearly mentions training on your public posts.