Can You Keep Your Data Private While Using AI? Here’s What Proton’s CEO Says

Introduction

Proton CEO Andy Yen has built a company whose entire pitch is privacy. ProtonMail, ProtonVPN, and Proton Drive all rely on end-to-end encryption and a business model that avoids selling user data. So when Yen says something keeps him up at night about artificial intelligence, it’s worth listening.

The concern isn’t that AI is inherently evil. It’s that the way most people interact with AI today—through free chatbots, image generators, and voice assistants—creates a leaky data environment. Every prompt you type, every file you upload, and every conversation you paste into a tool like ChatGPT or Copilot can be used to train the model, shared with third parties, or stored in ways you didn’t intend.

But Yen also believes privacy in the AI era is possible. The key is understanding where the risks are and making deliberate choices.

What Happened

In a recent interview with Spiceworks, Yen discussed the state of digital privacy as AI tools become ubiquitous. He pointed out that the most immediate threat isn’t a government subpoena or a sophisticated hack. It’s the routine, voluntary surrender of personal data to AI services that have ambiguous privacy policies.

For example, many popular AI tools log and review your conversations to improve their models. Some store data on servers in jurisdictions with weak privacy laws. Others reserve the right to share aggregated data with advertisers. Yen noted that users often assume their data is protected simply because a company says it cares about privacy, but the fine print tells a different story.

Why It Matters

The stakes are higher than most people realize. When you use an AI assistant to draft an email, summarize a confidential document, or brainstorm personal ideas, you’re feeding the model information that could be reconstructed later—even if the service promises to anonymize it. Research has shown that anonymization is notoriously difficult to guarantee, especially with large language models that can memorize training data.

For consumers, this means that sensitive details about health, finances, relationships, or work could end up stored or repurposed in ways you never agreed to. For professionals, the risk is even greater: a lawyer pasting client details into a public chatbot, a journalist using an AI tool to transcribe interviews, or a developer sharing proprietary code. All could face serious consequences if that data surfaces.

Yen’s specific worry, as he put it, is that people are giving away their data without understanding the trade-offs. And once that data is out there, it’s very hard to get back.

What Readers Can Do

The good news is that you don’t have to stop using AI to protect your privacy. You just need to be more intentional. Here are practical steps you can take today:

  • Choose services that take privacy seriously. Proton now offers Proton AI, which runs on its own encrypted infrastructure. Other options include tools that run locally on your device, such as llama.cpp or GPT4All, where data never leaves your machine. Before signing up for any AI service, read the privacy policy—specifically sections on data retention and third-party sharing.

  • Limit what you share. Even with a trusted provider, avoid typing sensitive information into a prompt. If you need to summarize a confidential document, use a local tool or redact identifying details first. Treat every AI interaction as something you wouldn’t mind seeing posted online.

  • Use pseudonyms and dummy account details. When creating accounts for AI services that are not privacy-focused, use an email alias (like from Proton Mail or SimpleLogin) and avoid giving your real name, phone number, or other identifiable information.

  • Review settings carefully. Many AI tools have a history or conversation log turned on by default. Turn it off if possible. Some services allow you to delete your conversation history; do this regularly. For browsers that integrate AI, check the privacy settings to prevent data from being sent to the vendor.

  • Consider the trade-off between quality and privacy. Free AI services often monetize through data. Paid subscriptions can reduce that incentive, but not always. Proton’s model is subscription-based without ads, and they are transparent about their code being open source. That kind of transparency is a good sign.

Sources

  • Spiceworks, “Privacy in the AI era is possible, says Proton’s CEO, but one thing keeps him up at night,” June 4, 2026.
  • Proton official website (proton.me) for details on services and encryption.
  • Public statements from Andy Yen on AI privacy, including interviews and blog posts.