Privacy in the AI Era: Practical Advice from Proton’s CEO
Introduction
Every time you paste a private note into a chatbot or ask an AI assistant to draft a sensitive email, you’re handing a piece of your life to a company you likely know little about. The trade‑off between AI convenience and privacy feels unavoidable. Proton, the company behind encrypted email and VPN services, argues otherwise. In a recent interview with Spiceworks, CEO Andy Yen said privacy in the AI era is possible—but one particular threat still worries him. Here’s what everyday users can learn from his perspective.
What Happened
Proton has built its reputation on end‑to‑end encryption and a zero‑access architecture: even Proton cannot read users’ emails or files. As AI tools proliferate—many of them free, cloud‑based, and hungry for data—Yen warned that similar standards are needed for AI interactions. The interview, published in June 2026, outlined Proton’s approach to offering AI features that respect user privacy, and highlighted a specific risk that keeps him up at night.
Why It Matters
Most popular AI services today (including chatbots, writing assistants, and image generators) process user input on remote servers. That data can be used to train models, sold to third parties, or leaked during a breach. For privacy‑conscious users, the stakes are real: sensitive business conversations, medical questions, or even personal stories entered into an AI tool become part of a corporate data set unless you take explicit steps to opt out or choose a privacy‑focused alternative.
The core issue is that many people assume their conversations are private, while the fine print often says the opposite. Proton’s CEO pointed out that even when companies promise not to train on your data, the data still sits on servers outside your control. Encryption—especially end‑to‑end encryption—is the only way to ensure that neither the provider nor a government can read your AI inputs.
What Readers Can Do
Drawing from Yen’s comments and Proton’s own practices, here are three practical steps you can take right now to reduce your data exposure when using AI.
1. Use On‑Device AI When Possible
If you only need basic tasks like summarization or text completion, look for AI models that run locally on your phone or computer. Apple’s on‑device models (available on recent iPhones and Macs) and many open‑source alternatives never send your data to the cloud. For drafting emails or notes, this keeps your text entirely within your control.
2. Opt Out of Model Training
Every major AI service—ChatGPT, Claude, Gemini—allows you to disable training on your conversations. This option is often buried in settings or requires a paid subscription. Go into your account preferences and turn it off. Even if you still send data to the cloud, this prevents it from being reused to train future versions of the model. It’s not perfect, but it’s a necessary first step.
3. Choose Encrypted AI Tools
Proton itself plans to offer AI services that use the same zero‑access encryption it applies to email. Several smaller companies already provide encrypted AI chat clients. Look for products that advertise “end‑to‑end encryption” specifically for AI conversations. If a tool cannot explain how it protects your data from its own staff, assume it doesn’t.
The One Thing That Worries Proton’s CEO
Yen pointed to a growing threat that even privacy‑conscious users overlook: AI chat logs that contain enough personal detail to enable identity theft. A single conversation about your travel plans, health history, or family relationships can, when pieced together, be used to answer security questions or impersonate you. His advice: treat AI chat logs as if they were passwords. If you wouldn’t paste your passport number into a public forum, don’t paste it into an AI assistant. Delete conversations that contain sensitive information, and use services that offer automatic deletion after a set period.
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 blog and documentation on zero‑access architecture and encrypted services.
You don’t have to stop using AI to protect your privacy. But you do have to be deliberate—choose tools that encrypt your data, turn off training, and think twice before sharing anything you wouldn’t want a stranger to read. Proton’s CEO is right: the technology exists. The hard part is choosing to use it.