For many, the rapid ascent of AI personal assistants like ChatGPT has raised significant privacy alarms. These powerful tools often necessitate sharing personal information, which is then retained by their parent companies. With OpenAI already exploring advertising models, concerns are mounting that the extensive data collection practices seen with platforms like Facebook and Google could soon extend to intimate chatbot interactions.
Addressing these concerns, Moxie Marlinspike, the visionary co-founder behind the encrypted messaging app Signal, has unveiled a new project called Confer. Launched in December, Confer demonstrates what a truly privacy-conscious AI service can look like, offering a compelling alternative to mainstream AI assistants.
Confer is meticulously designed to mirror the user experience of popular chatbots like ChatGPT or Claude. However, its backend is fundamentally structured to prevent data collection, leveraging the same open-source rigor that has made Signal a trusted name in secure communication. Crucially, your Confer conversations cannot be used to train the underlying AI model or to target advertisements, simply because the host never gains access to their content.
“It’s a form of technology that actively invites confession,” says Marlinspike. “Chat interfaces like ChatGPT know more about people than any other technology before. When you combine that with advertising, it’s like someone paying your therapist to convince you to buy something.”
Achieving this level of privacy necessitates a sophisticated interplay of several distinct systems.
Advanced Encryption and Secure Processing
First, Confer encrypts all messages to and from the system using the WebAuthn passkey system. While this standard performs optimally on mobile devices or Macs running Sequoia, it can also be configured for Windows or Linux using a password manager.
On the server side, all of Confer’s inference processing—the act of generating responses to user queries—is performed within a Trusted Execution Environment (TEE). These TEEs are fortified with remote attestation systems, which continuously verify that the system has not been compromised. Inside this secure environment, an array of open-weight foundation models handles incoming queries.
This intricate setup is considerably more complex than a standard inference architecture, which is already quite sophisticated. However, it is precisely this complexity that allows Confer to deliver on its fundamental promise: users can engage in sensitive conversations with the AI model, confident that their information remains private and secure.
Pricing and Accessibility
Confer offers a free tier, limited to 20 messages daily and five active chats. For unlimited access, along with more advanced models and personalization features, users can subscribe for $35 per month. While this price point is notably higher than ChatGPT’s Plus plan, it underscores the premium associated with robust digital privacy.
