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.

Used: News Factory APP - news discovery and automation - ChatGPT for Business