Feeling Unsafe Talking to AI? Signal Launches Secure Chatbot That Even the Company Can’t Read

Moxie Marlinspike, founder of Signal, has launched Confer, the world’s first fully secure AI chatbot. Built on end-to-end encryption and Trusted Execution Environment (TEE) technology, Confer ensures that not even the company, government, or hackers can access your chats. This open-source platform, which operates without passwords or email accounts, promises to be a game-changer for privacy-conscious users.

Complete Privacy for Users
AI chatbots can be intimidating, especially with fears that conversations might be leaked or monitored. Confer addresses this concern by keeping all chats completely secure, even on the company’s servers. Marlinspike explains that Confer is open-source, meaning anyone can inspect the code to understand how it works. Users log in using a PassKey, not a traditional password, and the server does not store a copy. This ensures that no one—not the company, hackers, or government agencies—can access the chat data.

Data Security on Servers
Unlike conventional systems where server administrators can view user data, Confer employs TEE technology—a secure “room” where no external party can peek inside. This guarantees complete protection of conversations, even at the server level.

Signal-Like Experience Without Compromising Privacy
Confer functions similarly to Signal in terms of privacy: users are not linked to email addresses or IP addresses. Chats can sync across multiple devices without compromising privacy. Marlinspike notes that many users feel comfortable sharing sensitive information on Confer—something they would hesitate to do on platforms like ChatGPT due to fears of data leakage.

A Challenge for Major AI Platforms
Currently, major AI platforms like Google, OpenAI, and Microsoft do not provide end-to-end encryption for user chats, and there are no indications of such updates in the near future. Legal and business constraints often prevent these companies from offering complete privacy. Experts suggest that the emergence of models like Confer could increase pressure on large AI providers to enhance user privacy and security.