Privacy Is Possible in the AI Era, Says Proton CEO – But Here’s What Worries Him

Using AI tools today usually means trading a bit of privacy for convenience. Every query you type into ChatGPT, Gemini, or Copilot travels through servers that may store, analyze, or even train models on your data. For many users, that trade-off feels unavoidable. But Proton CEO Andy Yen argues it doesn’t have to be that way. In a recent interview with Spiceworks, Yen laid out how privacy can coexist with AI — and why one particular threat keeps him up at night.

What happened

The interview, published June 4, 2026, covered Proton’s approach to AI privacy. Proton, best known for its encrypted email, VPN, and cloud storage, has been adding AI features to its own products. Yen stated that the company believes end-to-end encryption and on-device processing can make AI both useful and private. He stressed that the core issue is not the technology itself but how companies choose to deploy it.

Yen’s biggest concern, however, is the lack of transparency in how most AI providers handle user data. Many services collect enormous amounts of personal information — not just the content of your queries but also metadata like timestamps, device fingerprints, and behavioral patterns. This data can be aggregated, analyzed, and potentially shared with third parties or used to train future models without meaningful disclosure. Yen called this the “data aggregation risk” and said it represents the most serious threat to consumer privacy in the AI era.

Why it matters

The worry is not hypothetical. When you paste a draft email or ask a personal health question into a chatbot, you are effectively handing over sensitive data to an opaque system. Even if a company promises not to misuse that data, you have no way to verify what happens behind the scenes. And because AI training requires vast datasets, there is a strong incentive for providers to hold onto and reuse whatever users input.

For Proton, the answer is to design AI features that never send raw data to outside servers. The company’s approach relies on local processing when possible and end-to-end encryption when cloud processing is necessary. Yen believes this should become an industry standard, not just a niche offering. The interview also highlighted that Proton’s own AI features — such as smart replies or email summarization — are built on this privacy-first architecture, meaning users keep control over their data.

What readers can do

You don’t need to switch all your tools overnight. But you can take concrete steps to limit how much of your personal life ends up in AI training pipelines.

  • Use privacy-focused AI services where available. Alternatives like Proton, or services that clearly state they do not log or retain queries, give you more control. For general-purpose AI, check the privacy policy for phrases like “we do not train on your data” or “data encrypted at rest and in transit.”
  • Avoid sharing sensitive information with any AI tool. This includes full names, addresses, financial details, medical history, passwords, or anything you would not want publicly associated with you. Treat a chatbot like a stranger on the internet — even if the conversation feels private.
  • Review the privacy policy before clicking “accept.” Many AI services change their data practices over time. Yen advised users to look specifically for clauses on data retention, third-party sharing, and whether the service uses submitted data for model training. If the policy is vague, consider that a red flag.
  • Turn off chat history or training options. Most major AI platforms offer a setting to disable storing conversations. It is often buried in account settings, but it reduces the risk of your queries being used later. Note that this may limit some features, but privacy is worth the trade-off.
  • Be skeptical of “free” AI tools. If you are not paying with money, you are likely paying with your data. Consider whether a paid, transparent alternative fits your budget and needs.

Sources

The interview with Proton CEO Andy Yen was published by Spiceworks on June 4, 2026. You can read the full article at the link below (behind a soft paywall). For further background, Proton’s own website documents its encryption practices and the architecture of its AI features.

  • 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 site: Security and privacy model for AI features