Moxie Marlinspike, the brains behind Signal, is tackling a new frontier: AI privacy. He's been quietly developing Confer, an open-source AI chatbot with end-to-end encryption. His goal? To give us a space where our AI conversations remain truly private.

While everyone's jumping on the AI bandwagon, Marlinspike is wary. He's not thrilled about how easily our personal data can be accessed and used on most AI platforms. Think about it: when you chat with something like ChatGPT, it feels like a safe space. But behind the scenes, companies are potentially mining your words.

Marlinspike points out that we share intimate details with these AI, revealing our thoughts and uncertainties. This kind of info could be a goldmine for advertisers or even be used to manipulate us. I mean, who wants their deepest thoughts used to sell them stuff?

Confer is designed to encrypt both your questions and the AI's responses, ensuring that only you can access them. Imagine being able to explore your thoughts without the fear of them being used against you! Confer is all about breaking that cycle. It's a place where you can learn without becoming a product for data brokers.

Just like Signal, Confer encrypts your conversations right on your device before they even reach the servers. When the AI responds, it uses something called "confidential computing." This is where code runs in a secure environment that even the host machine can't access. It's like having a super-secure room for the AI to "think" in.

What I find really neat is that the hardware provides proof that everything's running securely. So, your device can verify that your data isn't being siphoned off to some giant data lake. It’s all about making sure your data stays yours.

Signal's messaging protocol was eventually adopted by WhatsApp. So, who knows? Maybe Meta and other big tech companies will eventually embrace Confer's technology too. One can hope! This could be a game-changer for how we think about privacy in the age of AI.