Privacy-Focused AI Assistant 'Confer' Offers Secure Conversations

A new project, Confer, initiated by Signal co-founder Moxie Marlinspike, aims to provide a privacy-conscious AI personal assistant. The project was launched in December and is designed to function like popular chatbots such as ChatGPT while ensuring user data security. Here are some key details:

• Concerns with AI Assistants: The rise of AI personal assistants often includes privacy risks, as user data is commonly collected by companies for various purposes, including advertising. This concern is evident with OpenAI testing advertising integration.
• Privacy-first Design: Confer's backend is designed to avoid data collection. The service employs open-source principles, similar to those seen in Signal, ensuring that user conversations cannot be accessed for model training or advertisement targeting.
• Technical Features: Confer encrypts communications using the WebAuthn passkey system and processes data in a Trusted Execution Environment (TEE) with remote attestation systems. This setup ensures data privacy even though it is more complex than standard arrangements.
• User Access: Confer offers a free tier with limited use, while a paid plan costing $35 monthly provides unlimited access and enhanced features. This service is priced higher than ChatGPT’s Plus plan, reflecting its focus on privacy.
• Marlinspike’s Perspective: Marlinspike emphasizes the intimate interactions people have with AI assistants and the potential for misuse if advertising were introduced. Confer's privacy mechanisms aim to prevent users from being manipulated through their personal data.

Overall, Confer represents a significant innovation in privacy-centric AI technology.