How to Protect Your Privacy When Using AI: Lessons from Proton’s CEO

Introduction

Every week there’s a new AI tool promising to save time, write emails, generate images, or summarize documents. But for anyone who’s thought twice before pasting personal information into a chat window, the privacy question is real: what happens to the data you share? Proton’s CEO Andy Yen recently told Spiceworks that privacy in the AI era is possible, but he also highlighted one specific risk that keeps him up at night. That risk, and what you can do about it, is worth understanding.

What happened

In an interview with Spiceworks, Yen laid out Proton’s view on AI and privacy. He acknowledged that AI services can be genuinely useful, but he pointed to a central danger: many AI providers collect user inputs and use them to train their models. This practice is often buried in terms of service, and unless you opt out — assuming you even can — your questions, drafts, and personal data become part of a model that other users will query. Yen said this kind of data leakage is the thing that most concerns him.

Proton, which operates from Switzerland under strong privacy laws, takes a different approach. Their recently launched Proton Scribe, an AI writing assistant, is designed so that all processing happens locally on the user’s device. No data leaves your computer, and no inputs are available for model training. For cloud‑based AI features in Proton’s services, the company uses a zero‑access architecture where even Proton cannot read your content.

Why it matters

The default for most free and many paid AI tools is that your data is the product. When you ask a chatbot to “improve this cover letter” or “summarize my meeting notes,” you are feeding it information that you might not want shared with others or stored indefinitely. Once data is used for training, it can be very difficult to remove. Even if a company promises not to train on your inputs, you have to take their word for it unless the system is technically auditable.

Yen’s point is that privacy in the AI era is not impossible, but it requires deliberate design, not just a promise. The “one thing” that bothers him is not AI itself — it’s the lack of transparency and user control over personal data in the rush to build smarter models. For consumers and professionals, this means the decision of which AI tool to use has a privacy cost that is easy to overlook until it’s too late.

What readers can do

You don’t need to abandon AI to protect your privacy. Here are concrete steps that apply whether you use Proton’s services or not:

  1. Check whether inputs are used for training. Look at the privacy policy and terms of service for any AI tool you use. Some let you opt out, but this setting is often hidden. If you cannot find a clear statement that your data will not be used for training, assume it will.

  2. Use end‑to‑end encrypted AI services where possible. Services that process data locally on your device (like Proton Scribe or some offline language models) prevent your data from ever reaching a cloud server. If you must use cloud‑based AI, look for providers that offer zero‑access encryption.

  3. Avoid sharing personally identifiable information in AI prompts. Even with good privacy protections, it’s good practice to redact names, addresses, account numbers, and other sensitive details. You never know how a prompt might be logged or stored.

  4. Enable privacy settings. Many AI platforms have a setting to turn off training on your conversations. For example, some allow you to disable “improve our models” options. Check your account settings regularly, as defaults can change.

  5. Consider using a VPN and encrypted storage. While not specific to AI, tools like a VPN (e.g., ProtonVPN) can mask your IP address when using cloud AI services, and encrypted cloud storage (e.g., ProtonDrive) keeps your documents safe if you let an AI assistant access them.

  6. Stay informed. The privacy landscape for AI is evolving quickly. What is safe today may change tomorrow as companies update their terms or as new threats emerge. Follow reliable sources like Spiceworks or technology‑focused privacy advocates.

Sources

The insights from Proton’s CEO are based on the Spiceworks article: “Privacy in the AI era is possible, says Proton’s CEO, but one thing keeps him up at night” (published June 4, 2026). You can read the full interview for more details on how Proton implements its privacy‑by‑design approach and what Yen sees as the biggest challenge ahead.