AI Privacy Is Possible, but Here’s What Still Worries Proton’s CEO

Using AI tools does not have to mean giving up your privacy—that is the message from Andy Yen, CEO of Proton, the company behind privacy-focused email, VPN, and cloud storage. In a recent interview with Spiceworks, Yen acknowledged that private AI is achievable, but he also pointed to a specific risk that keeps him up at night.

What happened

Yen, whose company has built a reputation on end-to-end encryption and zero-access architecture, spoke candidly about the tension between convenience and data protection in the age of generative AI. According to the interview, Yen’s biggest concern is not the AI models themselves but how user data is collected, stored, and potentially reused during interactions with AI services. Many popular chatbots and AI assistants send user prompts to cloud servers for processing, where they may be logged, analyzed, or even used to train future models—often without clear consent or meaningful opt-out options.

Proton has itself introduced AI features, such as Proton Scribe, which assists with email drafting. But the company has designed these tools to run locally or with strong encryption, so that user data never leaves the device or is decipherable by the service provider. Yen’s worry is that most consumers do not realize how much of their sensitive information—personal conversations, financial details, medical questions—is being fed into systems they do not control.

Why it matters

The concern is not hypothetical. Several AI companies have faced scrutiny or legal action over how they handle user data. ChatGPT’s parent company, for instance, has acknowledged using conversations for training unless users opt out. Other services have been caught storing prompts in plaintext or sharing data with third parties. For everyday consumers, the risk is that a private question about health, finances, or relationships could end up in a training dataset, or worse, be exposed in a breach.

Yen’s point is that the privacy threat from AI is not about the technology itself, but about the default settings and business models of many AI providers. “The thing that keeps me up at night is the assumption that because AI is useful, it must be safe,” he reportedly said. The real danger is that people hand over sensitive data without understanding the trade-offs.

What readers can do

The good news is that there are practical steps you can take to protect your privacy while still benefiting from AI tools.

1. Check the privacy policy of any AI service you use. Look for language about data retention, sharing with third parties, and whether prompts are used for training. Services that offer a clear “do not train on my data” option are better, but the most privacy-friendly ones process requests locally.

2. Prefer end-to-end encrypted AI tools. Proton Scribe, for example, runs on your device and never sends your email content to a remote server for analysis. Apple’s on‑device AI features similarly keep data local whenever possible. When a tool needs cloud processing, ensure the provider uses end-to-end encryption so that even the company cannot read your prompts.

3. Avoid sharing personal information in AI prompts when possible. If you need to ask a health‑related question, generalize the details. For financial or legal queries, consider whether the risk of disclosure outweighs the convenience.

4. Turn off chat history and training in your AI apps. Many popular chatbots allow you to disable history or opt out of model training. Do this in the settings menu before you start any sensitive conversation.

5. Use a privacy-focused browser or VPN when accessing AI services. This prevents your IP address and metadata from being tied to your prompts, adding another layer of separation.

6. Consider open-source or self‑hosted alternatives. Tools like Llama or Mistral can run on your own hardware, giving you full control over your data. This requires some technical comfort, but it is the most private option.

Sources

• Spiceworks interview with Proton CEO Andy Yen: “Privacy in the AI era is possible, says Proton’s CEO, but one thing keeps him up at night” (June 4, 2026)
• Proton’s official blog and product pages for Scribe (proton.me)
• General knowledge about AI data handling practices (e.g., OpenAI’s data usage policies, Apple’s on‑device processing)

The bottom line: Yes, you can use AI and still maintain privacy. But it requires you to be an active participant in protecting your data—not just a passive user. As Yen’s remark suggests, the technology is not the enemy; the ignorance of how it works is. Being informed is your best defense.