How to Protect Your Privacy While Using AI, According to Proton’s CEO

AI tools have become a daily habit for millions—writing drafts, summarizing articles, generating images. But every time you type a prompt into a chatbot, you’re sending that data to a company’s servers. For many people, that trade-off between convenience and privacy is starting to feel uneasy.

Recently, Proton’s CEO Andy Yen gave an interview where he acknowledged that privacy in the AI era is possible, but one particular concern keeps him up at night. His remarks offer a clear-headed look at what everyday users can do to protect themselves without abandoning useful AI tools.

What happened

In an interview with Spiceworks, Yen discussed Proton’s approach to AI and the broader privacy landscape. He expressed confidence that end-to-end encryption and privacy-first design can work with AI—Proton has already started integrating AI features into its encrypted email and drive services. However, he singled out one thing that worries him most: the centralization of data by a handful of large AI providers. When millions of users feed personal queries into a single model, that data becomes a tempting target for breaches, surveillance, or misuse.

Yen noted that many popular AI services collect user inputs to train their models, often without clear opt‑out mechanisms. Even when companies promise not to misuse data, the sheer volume of information being aggregated creates systemic risk.

Why it matters

The convenience of AI is real, but so are the risks. Every conversation with a chatbot, every uploaded document, every voice command can become part of a permanent record that a company retains. Even if a service claims to anonymize data, anonymization can often be reversed, especially when combined with other data sets.

For everyday users, the problem isn’t just about individual privacy—it’s about power. When a few companies hold vast amounts of personal data, they control how it’s used and who gets access. That’s the core of what keeps privacy advocates and even CEOs like Yen concerned.

What you can do

You don’t have to stop using AI, but you can make smarter choices about which tools you trust and how you use them. Here are practical steps based on what Proton’s CEO emphasizes and what privacy experts generally recommend.

  1. Prefer services that do not store your prompts.
    Look for AI tools that run locally on your device—like some models that can be downloaded and used offline. For cloud-based services, check whether they retain your conversation history. Services that only process data in memory and do not log inputs are better.

  2. Check the privacy policy for data usage in training.
    Many AI companies use your inputs to improve their models. This is often buried in the fine print. If you see language about “using data to improve services,” assume your prompts are being kept. Some services offer an opt-out (e.g., ChatGPT lets you disable training in settings), but not all do.

  3. Avoid sharing sensitive personal information in prompts.
    Obvious but worth repeating: don’t paste your passwords, financial details, or medical records into a chatbot. Even in encrypted services, the output might be stored or used elsewhere.

  4. Choose providers that use end-to-end encryption where possible.
    Proton and a few others are building AI features that run on encrypted data, so even the company cannot see your content. This is still rare, but it’s the gold standard.

  5. Use separate accounts or ephemeral sessions for non‑essential queries.
    If you don’t need a permanent chat history, use incognito modes or temporary accounts. Some services allow you to delete conversations manually—set a reminder to do this regularly.

  6. Consider open‑source or on‑device AI.
    Models like Alpaca, Llama 2, or smaller variants can run on a modern laptop. They won’t be as powerful as GPT‑4, but for basic tasks like summarizing or drafting, they are surprisingly capable and keep everything on your machine.

Sources

The original comments from Proton’s CEO Andy Yen were reported by Spiceworks in their article “Privacy in the AI era is possible, says Proton’s CEO, but one thing keeps him up at night” (June 2026). Additional context on privacy‑first AI design can be found in Proton’s own documentation and in independent reviews of major AI platforms’ data practices.

This article is intended for informational purposes and does not constitute legal or security advice. Practices and policies of AI services change frequently; always verify the current terms before use.