Signal Founder Integrates Encrypted AI Platform with Meta
Key Points
- Moxie Marlinspike, Signal founder, is integrating his encrypted AI platform Confer with Meta AI.
- Confer aims to provide end‑to‑end encryption for AI chat conversations, protecting data from companies, hackers, subpoenas, and governments.
- The integration will keep Confer independent while its technology underpins Meta’s AI services.
- WhatsApp head Will Cathcart highlighted the need for private AI in personal and confidential contexts.
- Researchers Mallory Knodel and JP Aumasson praised the initiative as an important step for private AI chatbots.
- Confer’s technology is built on open‑weight models, offering privacy while working with Meta’s advanced closed models.
- The collaboration could set a new standard for encrypted AI interactions across major platforms.
Moxie Marlinspike, the creator of Signal and its encryption protocol, announced that his privacy‑focused AI platform, Confer, will incorporate its technology into Meta AI. The collaboration aims to provide end‑to‑end encryption for AI chat conversations, shielding user data from AI companies, hackers, subpoenas, and governments. Researchers praised the effort as an important step toward private AI chatbots, while noting that the solution is still evolving. The move could bring encrypted AI interactions to millions of users across Meta’s services.
Background
Moxie Marlinspike, known for creating the secure messaging app Signal and its widely used open‑source encryption protocol, introduced a new privacy‑centric AI platform called Confer earlier this year. The platform is designed to deliver the power of generative AI while protecting conversations with end‑to‑end encryption.
Integration with Meta AI
Marlinspike announced that Confer’s encryption technology will be integrated into Meta AI. He explained that current AI chat services lack encryption, allowing data to be accessed by AI companies, employees, hackers, subpoenas, and governments. By embedding Confer’s privacy layer, Meta AI conversations could become private, preventing the data from being used for model training or other purposes.
Collaboration Details
Marlinspike emphasized that Confer will remain independent of Meta even as its technology underpins Meta AI. He described the goal as giving users “the full power of AI along with the full privacy of an encrypted conversation.” The partnership builds on Marlinspike’s earlier work in 2016, when he helped roll out end‑to‑end encryption for WhatsApp, a Meta‑owned service that recently added a Meta AI chatbot.
Industry Reactions
Will Cathcart, head of WhatsApp, highlighted the importance of private AI for personal and confidential use cases. Cryptography researcher Mallory Knodel of New York University called the development “great for people using chatbots that use Meta AI to have confidentiality and privacy within that exchange,” noting that Meta would not be able to access chat data for training. JP Aumasson, chief security officer at Taurus, called Confer “probably the best private AI solution, all things considered,” while acknowledging that documentation and threat modeling are still limited.
Technical Considerations
Confer’s technology is built on open‑weight models, allowing users to benefit from a wide range of tasks while missing some capabilities of proprietary frontier models. The collaboration gives Marlinspike the opportunity to work directly with Meta’s closed‑source models, potentially combining the most private AI chat technology with the most capable AI models available.
Outlook
Researchers see the partnership as a significant step toward broader adoption of encrypted AI, though they stress that the solution is not yet perfect. The effort demonstrates how trusted computing concepts, dating back to the 1990s, can be applied to modern generative AI to protect user privacy. If successful, the integration could set a new standard for private AI interactions across major platforms.