LUMO AI: An alternative swiss to chatgpt that respect privacy

·

·

Lumo Icon
Privacy-first AI assistant

Lumo AI: Proton's private-by-design assistant for everyday and professional use

Lumo AI is Proton's answer to AI assistants that monetize your data:
it does not use your conversations to train models, runs on infrastructure in Europe, and integrates
with the Proton ecosystem (Mail, Drive, Calendar, VPN) with a security model inspired by zero-access encryption.
Its philosophy is that AI should be useful without turning you into a product.





0 ads & tracking




No training on your chats




EU-based hosting




Zero-access design

Privacy by default
Part of Proton suite
Open-model stack

Key reasons to choose Lumo AI


Conversations stay yours

Lumo AI is designed so that your prompts and responses are not used to train models or build advertising profiles. The goal is to provide assistance without turning your data into an asset for third parties.


European infrastructure and compliance

Proton runs Lumo on infrastructure located in Europe, with an emphasis on GDPR alignment, minimizing the exposure of data to more permissive jurisdictions and bulk surveillance practices.


Integrated with Proton tools

Lumo connects with Proton Mail, Drive and other services to summarize, draft, and analyze within a private workspace, avoiding the need to upload documents to separate AI platforms.

Main features




  • Advanced writing and editing: drafts emails, reports, articles, and summaries with a style that can be tuned to your audience.



  • Developer assistance: explains code, suggests snippets, and helps debug in multiple languages, leveraging open models optimized for programming tasks.



  • Document understanding: uploads and summarizes files, extracts key points, and generates outlines for documents stored locally or in Proton Drive.



  • No training on user data: Lumo does not repurpose your chats or documents to improve its models, reducing the risk of leaks through future model behavior.



  • Optional browsing: web access is off by default, and when enabled it uses privacy-respecting search, avoiding unnecessary third-party tracking.



  • Lumo for Business: team-wide AI with organizational controls, EU hosting, and alignment with compliance requirements for sensitive data.
Applied privacy-by-design to AI

Instead of bolting privacy on top of existing ad-tech infrastructure, Proton built Lumo as part of a secure suite: encryption, jurisdiction choices, and data-minimization measures are design constraints, not optional extras.


Limitations to be aware of

By prioritizing privacy and self-hosted models, some tasks may feel slightly less "frictionless" compared to assistants deeply embedded in ad ecosystems, and feature rollout can be more conservative.

Practical comparison

Aspect Lumo AI (Proton) ChatGPT Gemini Microsoft Copilot
Use of chats for training by default Not used for model training Varies by plan / settings Data can be used, with controls Enterprise variants limit training, consumer can differ
Primary business model Subscription, no ads Subscription + ecosystem integration Advertising + ecosystem Licensing + ecosystem
Hosting & jurisdiction EU-based Proton infrastructure Global cloud regions Global Google Cloud Global Microsoft Azure
Integration with private email/drive Deep integration with Proton Mail & Drive Limited to external connectors Tight with Google Workspace (non-E2E) Tight with Microsoft 365 (non-E2E)
Open-source components Based on open models and open clients Mostly closed-source Mostly closed-source Mostly closed-source
Account requirements Proton account (free or paid) OpenAI / provider account Google account Microsoft account / tenant
Focus Privacy-centric personal & business use General-purpose AI Consumer + Workspace Enterprise & productivity suite
Transparency about training data Clear separation of user data and training High-level documentation High-level documentation High-level documentation

User experience and design

Functional and focused interface

The Lumo interface is deliberately simple: a clean chat canvas, history of conversations, and quick access to documents. The aim is to mimic the familiarity of a messaging app while keeping cognitive load low and avoiding distracting elements such as feed-style prompts or upsell banners.

Performance with a privacy constraint

By running models on Proton-managed infrastructure, Lumo balances responsiveness with data locality. Everyday writing and research tasks feel fast, while heavy workloads such as long documents or complex code may take a bit longer than hyperscale cloud models, especially during peak use.

Workflows for individuals and teams

  • Personal knowledge base summarization by combining notes, emails, and Drive documents in one assistant.
  • Drafting contracts, briefings, and project documents with context pulled from internal files rather than public cloud drives.
  • Iterative refinement of content, with versions stored in encrypted Proton storage instead of external AI servers.

Privacy in day-to-day use

  • Reduced risk of cross-service profiling because there is no ad network or social graph to feed.
  • Protection against casual data leakage by keeping sensitive texts in an encrypted ecosystem with minimal external connectors.
  • Consistency across devices: desktop and mobile access to Lumo under the same Proton account with unified security policies.

How to get started with Lumo AI

  1. Create or sign in to a Proton account: from the official Proton website, use an existing account or register a new one.
  2. Choose a plan: start with the free tier to explore Lumo or upgrade to a Proton plan that includes higher limits and Lumo Plus features.
  3. Access Lumo: open Lumo from the Proton web interface or install the dedicated mobile app if available for your platform.
  4. Review privacy settings: confirm that chat history, web browsing, and data retention behave according to your internal policies or personal preferences.
  5. Connect Proton services: optionally allow Lumo to access Proton Mail and Drive so it can summarize threads, draft replies, and work with documents.
  6. Define usage patterns: for professional teams, set internal guidelines indicating which types of data can be processed by Lumo and which should remain off-limits.

Recommended good practices

  • Classify documents before sending them to Lumo, and restrict highly sensitive materials to sanitized or redacted versions.
  • Keep Proton apps and operating systems updated to benefit from the latest security hardening and privacy improvements.
  • Use strong authentication on your Proton account (2FA) to prevent unauthorized access to Lumo and your encrypted data.


Mistakes to avoid

  • Assuming privacy is absolute: even with Proton's protections, compromised endpoints or screenshots can still leak information.
  • Feeding proprietary or regulated data into Lumo without first coordinating with legal and compliance teams.
  • Copy-pasting sensitive threads from other, less private platforms instead of working directly within Proton services.

Use cases

Knowledge workers and freelancers

Lumo helps draft proposals, contracts, and client communications while keeping inboxes and documents within Proton Mail and Drive, reducing the number of external services that handle sensitive information.

Journalism, civil society, and advocacy

Using an AI assistant that is not attached to a surveillance advertising network lowers the risk that research queries, sources, or internal notes feed external profiling or unclear retention practices.

Teams and organizations

For companies that cannot upload internal documents to generic AI tools due to policy or regulation, Lumo for Business provides a structured way to use AI while staying within Proton's privacy guarantees.

Security and privacy: layers that matter

Security in AI assistants is more than model quality: it includes encryption, data residency, data minimization, and governance. Lumo is designed to be conservative across all these dimensions.


Encryption and zero-access philosophy

Chats and stored artifacts follow Proton's zero-access philosophy whenever possible: data is encrypted so that Proton cannot read the content, limiting the damage of potential server-side compromises.


Metadata kept in check

Lumo seeks to reduce exposed metadata by limiting what is logged about who uses the assistant, when, and for what, avoiding the construction of granular behavioral profiles.


Real limits

No assistant, however private, can protect against risks such as compromised endpoints, malicious browser extensions, or screenshots. Security depends on user habits, device hygiene, and organizational controls.

Advantages and considerations

Highlights

  • Strong privacy stance: chats are not reused for training or advertising.
  • European infrastructure and alignment with Proton's encryption model and policy framework.
  • Integration with Proton Mail, Drive, and other tools in a single, secure environment.
  • Use of open models and open clients that invite scrutiny and community feedback.

Things to keep in mind

  • Performance and feature set may lag behind the very latest proprietary mega-models in some edge cases.
  • Requires commitment to the Proton ecosystem to get the most value from integrations.
  • Cannot neutralize organizational risks such as poor access control, weak passwords, or unsafe endpoint practices.

Frequently asked questions

Does Lumo AI use my conversations to train its models?

No. Proton's design goal with Lumo is to keep user conversations separate from training data. Your prompts and responses are not used to improve the models.

Do I need a Proton subscription to use Lumo?

You can start with a free Proton account and limited Lumo usage. Higher limits and advanced features are available through Proton paid plans and Lumo Plus tiers.

Can my company use Lumo in a compliant way?

Lumo for Business is designed for organizations that need AI while respecting internal policies and regulation. It runs on Proton infrastructure in Europe and offers controls for administrators.

Does Lumo have web browsing enabled by default?

No. By default, Lumo works with your prompts and your Proton data. Web access, when available, is opt-in and uses privacy-focused search approaches.

Is Lumo completely open source?

Lumo relies heavily on open models and open-source clients, but not every piece of the infrastructure is fully open. Proton publishes technical information so users can understand what is auditable and what is not.

Lumo Icon
Conclusion

Conclusion

Lumo AI proposes a different route for generative AI: instead of maximizing data collection and integration with advertising stacks,
it anchors itself in Proton's long-standing focus on privacy and encryption. The result is an assistant that may trade a bit of raw
convenience or "magic" for much stronger guarantees around how your information is handled. For individuals and organizations
that want AI support without widening their exposure surface or feeding additional surveillance infrastructures, Lumo offers a coherent,
privacy-centric option inside an ecosystem designed for confidentiality.

Leave a Reply

Your email address will not be published. Required fields are marked *