Crypto-Libertarian Erik Voorhees Warns Your Chatbot Queries Are Not Safe
Crypto-libertarian Erik Voorhees, fresh off his passionate defense of crypto in the 'Gold vs. Bitcoin' debate recently hosted by ZeroHedge, spoke with crypto news site Unchained's Laura Shin and Venice's COO Teana Baker-Taylor about the alarming honeypots of user search history data on AI chatbot platforms that could potentially be harvested.
Venice is one of Voorhees' latest crypto ventures. It is a private, uncensorable, open-source competitor to OpenAI's ChatGPT, powered by a decentralized crypto network.
Around the 25-minute mark, Shin asked Voorhees about the risks that OpenAI, Anthropic, and or other big AI companies could be doing with user search data.
Voorhees responded:
"Great question that's really like the most important question. So status quo today is you're using anthropic or OpenAI chat - you send in your question and it goes to that company and they store it forever and it's attached to your identity - right so they know that Lura Shin asked this question - um and they know that the AI responded back to you and they know what that is - and not only do they know that that question that conversation but they know your entire history of all conversations that you asked yesterday last year tomorrow and 10 years from now. All of it is associated with your identity.
"In the best case that's not that big of a deal, but in reality, what it means is - all your information - and essentially like parts of your mind like think your intellectual inquiries that you pursue - the things you think - the things you want to debate - the questions you have about life - and like big topics um can be known by Third parties."
"Advertisers for example, like that's not a huge deal if an advertiser knows something about you. But what if like uh a government knows something about you. What if um what if the Biden Administration learns that you are like uh you know orchestrating uh Trump's re-election campaign. What is what is the pressure on uh Biden Administration and OpenAI to use that information for something that people would consider corrupt and dangerous."
"These are very slippery slope arguments, um and it does not really matter what, like you know Anthropic's privacy policy says. If they have your information it will be shared with other parties today or tomorrow and probably both. And you can never get it back. So um that's the status quo - uh for people that are comfortable with that like keep using those services. that's okay."
Voorhees then explained how Venice AI prides itself in user sovereignty: "But like Venice was like well, um let's make a service that's just as easy" OpenAI and others, "but instead of like spying on you and recording all your information and attaching it to your identity forever let's just not do that."
* * *
Here's the full interview: