Proton CEO: The biggest AI privacy risk (and how to protect yourself)

If you’ve used ChatGPT, Claude, or any other AI assistant this week, your conversations might not be as private as you think. That reality is exactly what keeps Proton CEO Andy Yen up at night.

In a recent interview with Spiceworks, Yen laid out the privacy challenges that come with the rapid adoption of AI tools — and what his company is doing about them. The message is clear: using AI doesn’t have to mean giving up your data, but you need to know where the risks actually are.

What happened

Yen told Spiceworks that the single biggest concern for him is how AI models are trained on user data without explicit consent. Many popular AI services log your inputs — questions, documents, personal details — and use them to improve their models. In some cases, that data is stored indefinitely, shared with third parties, or even reviewed by human operators.

“The one thing that keeps me up at night is the widespread lack of transparency around what happens to your data when you use AI,” Yen said in the interview, published June 4, 2026.

Proton has been working on its own encrypted AI assistant (often referred to as Proton.ai), which processes user queries locally on the device or within end-to-end encrypted environments. The company also offers transparent policies that state no user data is used for model training. Yen pointed out that privacy-focused AI is not just possible — it’s necessary for building trust.

Why it matters

For the average person, the appeal of AI tools is obvious: they save time, write emails, summarize documents, and answer questions. But every time you paste a confidential email, a draft contract, or even a personal journal entry into a free AI chatbot, you are handing that information to a company that may store, analyze, or sell it.

The risk isn’t theoretical. Data breaches at AI providers have already exposed user conversations. And because many AI models retain information after you’ve deleted a conversation, you may lose control of your data permanently.

Regulation is still catching up. As another Spiceworks article from the same period notes, “Nobody knows who regulates your AI yet.” This regulatory vacuum means companies largely self-report their data practices — and some are not as forthcoming as others.

What readers can do

You don’t have to stop using AI. But you should be careful about which tools you trust and how you use them. Here are practical steps that match what Proton’s CEO recommends:

1. Choose providers with clear privacy policies.
Look for explicit statements that your data will not be used for model training. If the policy is vague or buried in legalese, consider it a red flag.

2. Avoid sharing sensitive information.
Do not paste passwords, medical records, financial details, or anything you wouldn’t want publicly known into any AI tool — even if it claims to be private.

3. Use on-device AI when possible.
Local processing means your data never leaves your device. Some apps and browsers now offer on-device AI features. They may be less powerful, but they are far safer for confidential work.

4. Check for end-to-end encryption.
Encrypted AI services (like Proton’s) ensure that even the provider cannot read your conversations. This is the gold standard for privacy.

5. Review and delete your data regularly.
If you’ve used a mainstream AI service, check its data retention settings. Delete old conversations if the option exists, and request data deletion where possible.

6. Be wary of free services.
If you are not paying for the product, you are the product. Free AI tools often monetize user data by selling it or using it for training. Paid subscriptions with clear privacy guarantees are safer.

Sources

The primary source for this article is the Spiceworks interview with Proton CEO Andy Yen, published June 4, 2026, titled “Privacy in the AI era is possible, says Proton’s CEO, but one thing keeps him up at night.” Additional context on regulatory gaps comes from another Spiceworks piece, “Nobody knows who regulates your AI yet,” published May 20, 2026. Proton’s official product pages and privacy policies were also referenced for information about its encrypted AI assistant.

This article is for informational purposes only. Individual privacy risks vary based on the services you use. Always verify data handling practices directly with the provider before sharing sensitive information.