NEWS

AI Chatbots and the Challenge of Privacy

A 2026 01 15T012155.697

As artificial intelligence chatbots become increasingly integrated into daily business operations, concerns about privacy and data security are rising. Moxie Marlinspike, the founder of Signal, has turned his attention to these issues, developing a new AI chatbot called Confer that prioritizes end-to-end encryption and user privacy.

Why Privacy in AI Matters

Marlinspike, a known advocate for secure communications, has voiced concerns about how current AI platforms handle user data. Large language models, such as those powering popular chatbots, often collect and store user conversations. These exchanges, which can feel as private as journal entries or one-on-one chats, are sometimes accessible to the companies behind the platforms and may be used for further training or even targeted advertising.

Marlinspike argues that if a chatbot interface feels private, it should actually function that way. He emphasizes that AI chatbots are unique in how they encourage users to share personal thoughts, uncertainties, and thinking patterns. This information, if not properly protected, could be exploited by advertisers or other third parties.

Introducing Confer: A Secure AI Chatbot

To address these privacy concerns, Marlinspike has introduced Confer, an open-source AI chatbot designed with robust encryption. Confer encrypts both user prompts and AI responses, ensuring that only the user can access their conversations. This approach aims to break the cycle of user data being repurposed for targeted ads or future AI training.

  • End-to-end encryption for all conversations
  • Open-source architecture for transparency
  • No use of traditional passwords; instead, passkeys such as Face ID, Touch ID, or device PINs are used to derive encryption keys

How Confer Protects User Data

Confer employs a combination of advanced security tools to keep user data confidential. Conversations are encrypted on the user’s device before being sent to Confer’s servers. Once received, prompts are decrypted only within a secure data environment to generate responses. The AI’s inference process runs in a confidential virtual machine, isolated by hardware to prevent unauthorized access.

Additionally, Confer uses cryptographic proof, known as attestation, to allow users to verify that their data is being processed securely. This system is designed to prevent user information from being stored in data lakes or accessed by unauthorized parties.

Potential Impact on the Tech Industry

Signal’s encrypted messaging protocol, introduced in 2014, was eventually adopted by major platforms such as WhatsApp. Marlinspike’s approach with Confer could influence how other tech giants address privacy in AI applications. As business owners and startups increasingly rely on AI tools, the demand for secure, transparent solutions is likely to grow.

For entrepreneurs and decision-makers, understanding the privacy implications of AI chatbots is critical. Solutions like Confer offer a blueprint for balancing innovation with the need to protect sensitive business and personal information.

Read More From the NEWS desk