Generative AI

Lumo AI 2.0 speeds up processing without compromising data protection

Artificial intelligence is gradually becoming a staple in productivity tools, but its development raises a question that has become impossible to ignore: What happens to the data we share with these digital assistants? While most major AI players rely on collecting information to improve their models, Proton is pursuing a radically different strategy. With Lumo AI 2.0, the Swiss company is enhancing its AI assistant with new reasoning capabilities, a customizable memory, image generation, and improved web search—all while making privacy its key selling point.1

This new version, therefore, does not seek merely to compete with ChatGPT, Claude, or Gemini in terms of performance. Above all, it aims to demonstrate that it is possible to offer modern artificial intelligence without turning users’ conversations into raw data used to train models or feed advertising systems.

For several years now, Proton has been building a digital ecosystem centered on the protection of personal data. Known for Proton Mail, Proton Drive, Proton VPN, and Proton Calendar, the Swiss company is now applying this same philosophy to artificial intelligence.

With Lumo AI 2.0, privacy isn’t just an added feature—it’s the very foundation of the service. Unlike some chatbots that retain conversations to improve their models or personalize their services, Proton asserts that conversations remain protected thanks to a secure architecture and a policy that minimizes the collection of personal data as much as possible.1

This approach addresses a growing concern among both individuals and businesses. As AI assistants become capable of handling confidential documents, strategic information, or business communications, the issue of digital sovereignty is taking on an importance comparable to that of technical performance.

Lumo AI 2.0 does more than just improve privacy. Proton is also introducing several enhancements designed to make its assistant more effective in everyday use.

The model now features a more advanced reasoning process. When faced with a complex request, the assistant gradually breaks down the problem, analyzes several possible approaches, and then constructs a more structured response. This improvement notably enhances the quality of summaries, literature reviews, and technical explanations.

Another important new feature is persistent memory. When a user grants permission, Lumo AI 2.0 can store certain preferences in order to gradually tailor its responses. Unlike systems where this memory remains largely opaque, Proton allows users to view, edit, or delete the stored information at any time. This ensures that users retain control over what the assistant actually remembers.2

The interface is intentionally kept simple and user-friendly. Proton Mail and Proton Drive users will find a familiar environment, which makes it easy to integrate the assistant into their daily routines without having to juggle multiple accounts or platforms.

MSc in Data Management at aivancity
aivancity

MSc in Data Management

Lead the strategic and operational management of data: acquisition, analysis, and governance. RNCP Level 7 degree (equivalent to a 5-year post-secondary degree).

1 to 2 years — 5 years of post-secondary education Bachelor's degree through Master's degree Work-Study Program or Traditional Education Paris-Villejuif Campus
Learn more about the program → RNCP Level 7 Certification

Proton is also enhancing Lumo AI 2.0 with features that are gradually becoming essential in next-generation virtual assistants.

The assistant can now search the web to incorporate up-to-date information into its responses when necessary. This capability significantly improves results for current events, professional research, and literature reviews. Proton also cites the sources used to enhance the transparency of the responses it provides.1

Image generation is also making its debut. Users can quickly create illustrations for presentations, working documents, or digital publications directly from the Lumo AI 2.0 interface. Proton notes, however, that it has incorporated several mechanisms designed to limit misuse and ensure compliance with intellectual property rules.

All of these features are gradually transforming Lumo AI into a true productivity assistant rather than just a conversational chatbot.

The reason Proton places such a strong emphasis on privacy is that its assistant is aimed at an audience that is particularly sensitive to these issues.

Law firms, government agencies, consulting firms, and research teams handle strategic information on a daily basis that cannot be entrusted to just any cloud service. For these organizations, the ability to use artificial intelligence without exposing their data is now a significant competitive advantage.

The Swiss legal framework is another distinguishing feature. Proton benefits from legislation recognized for its high level of personal data protection, which builds trust among many European users compared to major U.S. platforms.

However, this philosophy does not mean that Lumo AI 2.0 sacrifices the user experience. On the contrary, Proton seeks to demonstrate that a high-performing assistant can function perfectly well without basing its business model on the exploitation of personal data.

Proton is gradually rolling out Lumo AI 2.0 to users in its ecosystem. The assistant is accessible directly from the Lumo web app, as well as through various compatible Proton services, without requiring any complex installation.3

European users, including those in France, can already access the service, as Proton is prioritizing a simultaneous global launch—unlike some other services, whose rollout remains limited to the United States for several weeks.

The company offers a free version that allows users to explore the assistant’s main features, while users of the Proton Unlimited and Proton Visionary plans enjoy expanded access with larger quotas, greater processing capabilities, and new advanced features such as persistent memory and image generation.3

For organizations looking to integrate the assistant into their business workflows, Proton is also continuing to develop its enterprise offering, with gradual integration across its entire secure ecosystem.

The launch of Lumo AI 2.0 shows that competition among AI assistants is no longer based solely on model size or benchmark scores.

OpenAI, Anthropic, Google, and Microsoft are investing heavily in performance and intelligent agents. Proton is taking a different approach: digital trust. The company believes that the next generation of assistants must not only be able to generate relevant responses but also ensure that the information entrusted to them remains under the control of its owners.

This strategy could appeal to a growing number of European organizations grappling with the requirements of the GDPR, digital sovereignty policies, or issues related to the confidentiality of business data.

The evolution of Lumo AI 2.0 illustrates a more profound transformation in the artificial intelligence industry. For a long time, users were primarily concerned with model performance. Now, the way these models handle personal data has become just as important.

The development of assistants capable of remembering preferences, analyzing confidential documents, or generating professional content requires much stricter governance of the information exchanged. In this context, companies will need to balance the range of features, ease of use, and level of protection offered by each provider.

Lumo AI 2.0 shows that a different approach is possible. Although no solution can guarantee zero risk, Proton has chosen to limit data collection right from the design phase of its assistant rather than adding protective measures after the fact.

With Lumo AI 2.0, Proton isn't just looking to launch a new chatbot. The company is proposing a different vision of artificial intelligence, one in which technical performance goes hand in hand with enhanced control over personal data.

As AI assistants become increasingly integrated into professional and personal environments, this approach could become a true differentiator. In a market dominated by a few major U.S. players, Proton points out that the future of artificial intelligence will depend as much on the trust placed in these platforms as on their technological capabilities.

Technology Framework

How does Lumo AI 2.0 work?

Lumo AI 2.0 is the artificial intelligence assistant developed by Proton, the Swiss company behind Proton Mail, Proton Drive, Proton VPN, and Proton Calendar. Unlike most AI assistants on the market—which are designed primarily to maximize performance or improve models using user data—Lumo AI 2.0 takes a privacy-first approach. Its architecture combines advanced language models, encryption mechanisms, and transparent data management to provide an environment where users retain control over their personal information.

The model relies on enhanced reasoning capabilities that allow it to break down complex requests into several logical steps before constructing a structured response. This approach improves the quality of summaries, literature reviews, research, and tasks requiring in-depth reasoning. Lumo AI 2.0 also includes an optional persistent memory feature. When the user grants permission, the assistant can store certain preferences to gradually personalize its responses. This information can be viewed, edited, or deleted at any time.

The model also features a built-in web search function that allows it to access recent information, enriching its responses with up-to-date content while citing the sources it consulted. This capability improves the relevance of responses on current events, professional research, or research projects. Proton has also added image generation directly within the assistant to enable the quick creation of illustrations for presentations, documents, or digital content.

All of these features are built on Proton’s secure infrastructure. Communication is governed by a strict data protection policy and an architecture designed to minimize the collection of personal information as much as possible. This philosophy sets Lumo AI 2.0 apart from many AI assistants whose business models rely in part on the exploitation of user data.

Key Features of Lumo AI 2.0
  • Advanced Reasoning: Analysis of Complex Requests Through a Logical, Multi-Step Breakdown
  • Controlled Persistent Memory: Optional customization of responses, with the ability to manage or delete stored information
  • Integrated web search: access to up-to-date information with sources displayed when necessary
  • Image Generation: Creating Illustrations Directly from the Conversational Interface
  • Secure Conversation Management: An Architecture Designed to Limit the Collection of Personal Data
  • Integration with the Proton ecosystem: compatibility with Proton Mail, Proton Drive, Proton Calendar, and other services on the platform
  • Swiss Infrastructure: Hosting and Governance in Compliance with the Strict Requirements of Swiss Data Protection Laws
Technical constraints and limitations
  • Some advanced features are still available only to paid Proton subscribers
  • Reasoning performance may vary depending on the complexity of the queries and the models used
  • Its image-generation capabilities are still more advanced than those of some specialized players in the market
  • Persistent memory requires voluntary activation by the user
  • Like all language models, Lumo AI 2.0 may produce factual errors or hallucinations and requires human validation for critical uses
  • Enhanced privacy does not eliminate the need to follow cybersecurity best practices, including multifactor authentication and account protection

The development of Lumo AI 2.0 illustrates a major evolution in conversational assistants, where performance can no longer be separated from issues of privacy and data protection. On a related topic, check out our article “Temporary Conversations on Gemini: A Step Toward More Privacy-Respectful AI”, which analyzes how major AI players are evolving their services to give users more control over their data and interactions.

1. Proton. (2026). Introducing Lumo AI 2.0.
https://proton.me/lumo

2. Proton Support. (2026). Using Memory and Privacy Controls in Lumo.
https://proton.me/support/lumo

3. Proton. (2026). Lumo AI Plans and Availability.
https://proton.me/lumo/pricing

Don't miss our upcoming articles!

Get the latest articles written by aivancity experts and professors delivered straight to your inbox.

We don't send spam! Please see our privacy policy for more information.

Don't miss our upcoming articles!

Get the latest articles written by aivancity experts and professors delivered straight to your inbox.

We don't send spam! Please see our privacy policy for more information.

Related posts
AI & EducationGenerative AI

ChatGPT for Teens: OpenAI Wants to Guide the Generation Growing Up with AI

On August 18, 2026, OpenAI launched a ChatGPT pilot program for 13- to 17-year-olds, featuring enhanced safeguards, age estimation, learning tools, and parental controls. The program is not a…
Generative AI

Gemini 3.7 Flash: Google Boosts Its AI for Coding and Automation

Just three weeks after Gemini 3.6 Flash, Google is already stepping up the pace with Gemini 3.7 Flash, a new model designed to boost performance in programming, automation, and workflows involving multiple AI agents…
Technological Advances in AIGenerative AI

GPT-5.6 Sol Accelerates 10x: OpenAI Unveils Its New Ultrafast Mode

A few weeks after rolling out GPT-5.6 on a large scale, OpenAI is tackling another major challenge in artificial intelligence: latency. With its new Ultrafast mode, the company aims to enable its flagship model, GPT-5.6 Sol, to produce…