Privacy in the AI Era Is Possible—But Here’s What Proton’s CEO Really Worries About

When Proton’s CEO Andy Yen says privacy and AI can coexist, people in the security world listen. Proton has built its reputation on end-to-end encrypted email, VPNs, and cloud storage, and its leader has been outspoken about the trade-offs consumers face when using free online services.

In a recent interview with Spiceworks, Yen laid out a mostly optimistic view: it is possible to use AI tools without surrendering your privacy. But he also pointed to one threat that keeps him up at night—and it’s not the usual worry about hackers or government surveillance.

What happened

On June 4, 2026, Spiceworks published an interview with Andy Yen. He discussed how Proton is approaching the AI era, including integrating AI features into its services while maintaining encryption. But when asked what worries him most, Yen pointed to the widespread practice of using user conversations to train AI models without clear, upfront consent.

This isn’t an abstract fear. Many popular AI assistants—ChatGPT, Gemini, Copilot—log prompts and responses. That data can be reviewed by human annotators, used to improve the model, or, in some cases, stored indefinitely. And even when companies claim to anonymize data, the line between “anonymous” and “personally identifiable” is often blurry when you’re sharing details like job titles, medical history, or even weekend plans.

Why it matters for you

If you’ve ever typed a draft email into ChatGPT, asked Gemini for tax advice, or used Copilot to summarize a sensitive document, you’ve potentially handed over data that could be linked back to you. The problem isn’t malice—it’s that the default settings on most free AI tools assume consent. You opt in by using the service, and the fine print often gives the company broad rights to use your input.

Think about it: a single conversation with a chatbot might contain your name, employer, financial situation, or health concerns. If that data feeds into a public-facing model or gets accessed by staff during training, you lose control over who sees it. Yen’s worry is that most users don’t realize this is happening until it’s too late.

What you can do about it

You don’t have to stop using AI tools to protect your privacy. But you do need to be intentional. Here are practical steps, inspired by the approach Proton advocates:

Be selective about what you share. Treat every prompt like a public post. If you wouldn’t put it on social media, don’t paste it into a free AI chat. Use generic examples instead of real personal data.

Check the privacy policy—then check again. Look for language about data used for training. Some services offer opt-out settings (like OpenAI’s “data controls”), but they’re often buried. Turn off chat history saving if possible.

Use services that prioritize privacy. Proton has started offering AI features that run on encrypted infrastructure. Other options include tools that process data locally on your device or use end-to-end encryption. For instance, Proton VPN and Proton Mail already protect your metadata; applying that philosophy to AI means choosing providers that don’t log or monetize your queries.

Consider a virtual private network when accessing AI tools. This prevents your IP address and location from being logged—a small step that reduces the digital footprint you leave behind.

Hold companies accountable. Ask your AI provider whether your data is used for training and how long it’s stored. If they can’t give a clear answer, that’s a red flag.

Sources

The interview with Andy Yen was published by Spiceworks on June 4, 2026. Proton’s product information is available on its official site. Additional context on AI data practices comes from public privacy policies of major AI providers and ongoing reporting by consumer privacy advocates.

The bottom line: Privacy in the AI era is possible, but it demands more than passive trust. As Yen’s comments remind us, the biggest risk isn’t the technology—it’s the assumption that someone else has your best interests in mind. A little skepticism goes a long way.