Proton launches Lumo 2.0 with image generation, memory, and private web search

Proton, best known for its encrypted Proton Mail service, announced Lumo 2.0, which it calls the largest upgrade to its privacy-first AI assistant to date. Lumo was built around five core principles: no logs, zero-access encryption, no data sharing, no use of conversations for training, and open-source language models.

Lumo 2.0 adds three major capabilities: multimodal image recognition and generation (analyze, edit, and generate images in one conversation, all under zero-access encryption); deeper context via user-controlled memory, encrypted Projects, and Custom Lumos; and enhanced web search with live results and source citations for more accurate, up-to-date answers.

Proton says more than 10 million people have adopted Lumo as a private alternative to leading AI platforms. It is also pushing Lumo for Business, targeting organizations that can't risk exposing sensitive data — every conversation is zero-access encrypted, never logged, and never used for training, with data hosted on independent European infrastructure that Proton says is beyond the reach of US Executive Orders and American data-collection requests.

The privacy-first positioning is a deliberate contrast to Google's Gemini, which just made personalized image generation free by reading users' Gmail, Photos, and YouTube data. That juxtaposition captures a genuine market split: convenience-through-data-access versus privacy-through-encryption. The trade-off for Lumo users is that open-source models and encryption constraints may lag frontier capabilities, but for privacy-sensitive individuals and European enterprises wary of US data jurisdiction, that's an acceptable price. Lumo's European infrastructure angle is a meaningful differentiator amid tightening data-sovereignty concerns.