Lumo “2.0” is the most significant upgrade since Proton’s initial AI assistant launch last year which had attracted over 10 million users. The new version rebuilds the platform on a fresh architecture that combines advanced reasoning capabilities with zero-access encryption which is a technical and philosophical departure from how mainstream AI services operate.
OpenAI’s ChatGPT, Anthropic’s Claude variants, and similar models have become essential tools for billions of users. Yet these services operate on a fundamental exchange: your data, conversations, and behavioral patterns become training material for future iterations. Queries are logged, analyzed and used to refine products sold back to you or monetized through advertising. For businesses, this creates acute risks: confidential documents uploaded to ChatGPT become part of corporate training datasets and employee conversations feed directly into models that competitors might later access.
© Gizmodo
Lumo 2.0 diverges sharply from this model as every conversation, image, and piece of context remains encrypted using the same zero-access architecture that protects over 100 million Proton Mail, Proton VPN and Proton Drive users: Neither Proton nor any third party can access the conversations. The data is never used to train the model. The infrastructure sits entirely in Switzerland and it is protected by Swiss privacy laws rather than subject to US Executive Orders or American data collection requests
