Proton CEO: The Biggest Privacy Risk in AI and How to Protect Yourself
Andy Yen, CEO of Proton, has said that privacy in the AI era is achievable—but he also pointed to one concern that keeps him up at night. In a recent interview cited by Spiceworks, Yen warned that the way most AI services centralize and use personal data poses a greater threat than many users realize. For people who rely on tools like ChatGPT, Grammarly, or image generators, the trade-off often comes down to convenience over control.
What Happened
Yen’s remarks, reported by Spiceworks, focused on the risks of data aggregation and unregulated model training. The core problem is that many AI platforms collect user inputs—conversations, documents, even private files—to improve their models. Once data is fed into a large language model, it becomes difficult to remove. And with most AI services operating on cloud servers, users have little visibility into how their information is stored or who can access it.
Proton, best known for its encrypted email and VPN services, has taken a different path. The company is building AI features that run on-device or use end-to-end encryption wherever possible. Their approach emphasizes anonymization and local processing, meaning your data doesn’t have to leave your device to be useful.
Why It Matters
The average consumer might not think twice before pasting a sensitive email or a draft contract into an AI assistant. But those interactions can be stored, analyzed, and used to train the very models that others will query later. Even if a company promises not to share your data, internal data breaches or policy changes can expose it.
The threat isn’t theoretical—several major AI services have faced scrutiny for using customer data without clear consent. And because AI models are built on vast datasets, removing a single user’s information is often impossible. Yen’s concern reflects a growing unease among privacy advocates: the convenience of AI is pulling users toward centralization, which weakens individual control.
For the tech-savvy consumer, the question is not whether to use AI, but how to do so without handing over more than necessary.
What Readers Can Do
You don’t have to stop using AI, but you can take practical steps to reduce your exposure. Here’s a straightforward checklist:
- Choose services that allow you to opt out of model training. Many tools now include a setting to prevent your data from being used for improvement. Enable it.
- Prefer on-device AI when possible. Some applications (like certain transcription tools or photo editors) can process data locally. This keeps your information off remote servers.
- Use end-to-end encrypted alternatives. Proton’s own tools are one example, but look for others that encrypt data in transit and at rest.
- Avoid pasting sensitive information into public AI interfaces. If you need help drafting a confidential document, consider a dedicated writing tool that doesn’t send text to the cloud.
- Read (or at least skim) privacy policies—especially sections on data retention and third-party access. Red flags include vague language, no opt-out, and long indefinite data storage.
- Review your account settings regularly. Providers sometimes change defaults, and you may need to re-opt out of training after updates.
If you’re evaluating a new AI tool, ask these questions: Does it work offline for core tasks? Can you delete your conversation history permanently? Is the company transparent about who has access to your data? If the answer to any is unclear, proceed with caution.
Sources
- Spiceworks: “Privacy in the AI era is possible, says Proton’s CEO, but one thing keeps him up at night” (June 2026) – summary and quotes cited above.
- Proton’s public documentation on AI features and encryption (proton.me).
Privacy in the AI era is possible, but it requires informed choices. Yen’s warning is a useful reminder: the real risk isn’t AI itself—it’s the unchecked collection of your data behind it.