The One AI Privacy Risk That Worries Proton’s CEO—And How to Protect Yourself

Intro

Using AI tools like ChatGPT, Gemini, or Microsoft Copilot has become routine for many of us. They help with writing, research, coding, and everyday tasks. But these conveniences come with a trade-off: the data you feed into them is often stored, analyzed, and sometimes used to train future models. In a recent interview with Spiceworks, Proton’s CEO Andy Yen shared what keeps him up at night when it comes to AI privacy—and it’s a risk many users overlook. Here’s what he said, and practical steps you can take today.

What Happened

In an interview published by Spiceworks, Andy Yen discussed the state of privacy in the AI era. While he remains optimistic that privacy is possible, he pointed to one persistent threat: the collection and retention of user conversations by AI providers. According to Yen, even when companies promise not to use your data for training, the lack of end-to-end encryption means your chats can still be read by the service provider—and potentially exposed in a breach or compelled by legal requests. He noted that this risk is especially acute for anyone handling sensitive work or personal information.

Why It Matters

Most major AI chatbots record your prompts, responses, and even metadata like timestamps and IP addresses. This data may be stored for months or years, and is sometimes reviewed by human moderators or used to improve models. Even if you never type anything flagrantly private, simple queries can reveal a lot about your life: what you work on, what you’re researching, health concerns, financial planning, and more.

The core issue is control. Once your data leaves your device and enters a corporate server, you lose the ability to know exactly what happens to it. Yen’s concern highlights a gap in the current AI ecosystem: users are often not told clearly how long their data is kept, who has access, or what steps are taken to protect it from third parties. For privacy-conscious consumers, this is not a theoretical problem—it’s a daily risk.

What Readers Can Do

You don’t have to stop using AI tools, but you can significantly reduce your exposure. Here are concrete steps:

1. Avoid sharing personal or sensitive information. Treat every AI conversation as if it could be read by a stranger. Never paste passwords, medical records, financial details, or confidential work documents. If you need to use AI for genuinely private tasks, consider more secure alternatives.

2. Use services that don’t log your chats. Some providers offer “incognito” modes or promise not to store conversations. For example, you can delete chat histories regularly, or use tools that run on-device, like certain local language models (e.g., Llama or Mistral on your own machine). These eliminate the need to send data to a cloud server.

3. Check and change your privacy settings. Most major chatbots allow you to opt out of having your data used for training. Look for settings labeled “do not train on my data” or “disable data retention.” Even with this enabled, the company will still see your chats in real time, but at least they won’t be stored long-term.

4. Explore encrypted or privacy-first AI tools. Proton, best known for its encrypted email and VPN, is exploring ways to bring the same principles to AI. While its full-featured AI offering is not yet widely available, other services like Brave’s AI (on-device) or privacy-focused front-ends (e.g., DuckDuckGo’s AI Chat with anonymization) are worth trying. They reroute queries or process them locally to reduce data exposure.

5. Use temporary sessions or browser-based isolation. When you must use a web AI tool, do so in a private browsing window and avoid logging into accounts that might link your identity to the chat. After finishing, close the session—this reduces the chance of cookies or persistent identifiers tying the conversation back to you.

Sources

  • Spiceworks, “Privacy in the AI era is possible, says Proton’s CEO, but one thing keeps him up at night” (June 2026). The interview with Andy Yen is the primary source for his concerns about data retention and lack of encryption in AI tools.

  • General research on AI data practices by the Electronic Frontier Foundation (EFF) and Consumer Reports further confirms that user queries are often stored and used for training, and that opt-out settings vary widely by provider.

No single step will make you invulnerable, but combining several of these habits goes a long way. As Yen pointed out, privacy in the AI era is possible—it just requires paying attention to where your data goes and taking a few deliberate precautions.