Proton CEO on the Biggest AI Privacy Threat and How to Protect Yourself

Proton CEO on the Biggest AI Privacy Threat and How to Protect Yourself

AI tools are now part of everyday life, from drafting emails to generating images. But each query you send could become training data for the next model update—or a record held by a third party. In a recent interview with Spiceworks, Proton’s CEO acknowledged that privacy in the AI era is achievable, but pointed to one concern that keeps him up at night: the erosion of user control over personal data as it flows into opaque AI pipelines.

What happened

Proton, best known for its encrypted email and VPN services, has been vocal about privacy since its founding. In the Spiceworks interview, the CEO discussed the tension between AI’s utility and the privacy trade-offs users often make without realizing it. While Proton is building encrypted AI features—such as on-device processing for its productivity tools—the CEO highlighted a broader systemic risk: AI models are being trained on massive datasets collected from users who have little say in how their information is used, combined, or retained.

He identified data aggregation and lack of transparency as the core threats. Even in consumer tools that claim to respect privacy, the underlying infrastructure can be a black box. When you paste a sensitive document into a chatbot or upload a photo for editing, you typically have no way of knowing whether that data will be stored, shared, or used to fine-tune future models.

Why it matters

The convenience of AI often comes with invisible costs. Most free and low-cost AI services are supported by data collection. The CEO’s worry is not just about shady actors—it’s about the normalization of surrendering control. Even well-meaning companies can inadvertently create privacy risks when their models are trained on user inputs that were never meant to be permanent.

Proton itself is trying to set a counter-example: it offers end-to-end encrypted AI features, such as smart compose and translation, that run locally on user devices. But the CEO acknowledged that individual companies can only do so much. The broader ecosystem—app stores, cloud providers, third-party APIs—still operates on a model that often values data volume over user consent.

For the average consumer, this means the tools you rely on might be collecting far more than you realize. And because AI training is a slow, cumulative process, the damage is not immediate. By the time a privacy breach or misuse is discovered, your data has already been absorbed into a system you cannot fully undo.

What readers can do

You don’t have to stop using AI to protect your privacy, but you can be more intentional about it. Here are a few practical steps, based on the concerns raised in the interview:

• Check the privacy policy of AI tools you use regularly. Look for language about data retention, third-party sharing, and whether your inputs are used for training. If the policy is vague or says “we may share anonymized data,” assume your information is being collected.
• Use services that process data locally. Proton’s encrypted AI features are one option, but others exist, such as local language models (LLMs) that run on your own device (e.g., Ollama, LM Studio). Even if you don’t switch completely, being aware of where processing happens matters.
• Avoid sharing sensitive personal information with general-purpose AI chatbots. Treat each conversation as if it could become public. That includes financial details, health information, passwords, or anything you wouldn’t want the world to see.
• Use separate, disposable accounts for AI tools when possible. Many services allow you to sign up with a masked email (like SimpleLogin, also owned by Proton) or a temporary email address. This limits the data trail tied to your real identity.
• Turn off chat history and model training in settings. ChatGPT, Google Gemini, and others offer options to disable storage of your conversations. It may slightly degrade personalization, but it reduces the chance your data will be used beyond your session.
• Support companies that are transparent about AI data practices. The Proton CEO’s message is that consumer demand drives change. When you choose a service that clearly states what it does with your inputs, you help shift the market toward better privacy norms.

Sources

• Spiceworks interview, “Privacy in the AI era is possible, says Proton’s CEO, but one thing keeps him up at night” (June 2026)
• Proton’s official blog and privacy documentation regarding encrypted AI features