As is typical, this science reporting isn’t great. It’s not only that AI can do it effectively, but that it can do it at scale. To quote the paper:
“Despite these models achieving near-expert human performance, they come at a fraction of the cost, requiring 100× less financial and 240× lower time investment than human labelers—making such privacy violations at scale possible for the first time.”
They also demonstrate how interacting with an AI model can quickly extract more private info without looking like it is. A game of 20 questions, except you don’t realize you’re playing.
A friend of mine explained why it’s important to his kids: they can’t chat with a group of their friends.
Why? Because parents don’t want to install WhatsApp or other group chats due to legitimate concerns about scammers, pedophiles, and other child predators. SMS chat fills that gap, but it breaks horribly for groups bigger than 10 people. Hence if some kid is on Android, they break their chat. Given the penetration of Apple devices, it’s the kids with Android who are considered at-fault. “Just get an iPhone!”
Welcome to anticompetitive practices targeted at your children.