Can You Actually Keep Your Privacy While Using AI? Proton’s CEO Weighs In

Every new wave of technology brings a familiar trade-off: convenience for privacy. AI tools are no exception. You ask a chatbot a question, let an app summarize your emails, or use an AI image generator — and in return, your data gets processed, stored, and possibly reused to train the next model. For many people, that exchange has started to feel less like a fair deal and more like a silent giveaway.

But is it possible to use AI without surrendering your privacy? According to Proton’s CEO, the answer is yes — with a few important caveats. In a recent interview with Spiceworks, he laid out what the company sees as the path forward and, more tellingly, what still keeps him up at night.

What happened

Proton, best known for its encrypted email and VPN services, has been building privacy-first AI tools for a while now. In the interview, the CEO argued that privacy in the AI era is achievable, but it requires the right combination of technology and policy. He pointed to end-to-end encryption and on-device processing as two technical foundations that can keep user data out of the hands of third parties.

But he also raised a specific concern: AI models, if not designed carefully, can inadvertently expose user data. Even when the infrastructure is secure, the model itself might regurgitate sensitive information it was trained on — or behavior that looks like a breach of trust. That, he said, is the one thing that worries him most. Not the technology itself, but the potential for a single misstep to erode the trust that privacy-conscious users have placed in these services.

Why it matters

The stakes here go beyond Proton. AI tools are already embedded in daily life — from writing assistants to customer service chatbots to search engines. Most of these are run by companies that collect and store data by default. Users rarely know exactly what data is being collected, how long it is kept, or whether it is used to train future models. And with the rapid adoption of generative AI, that opacity is only growing.

Proton’s CEO framed this as a turning point. If the industry continues to treat user data as a free resource, public trust will erode further. That hurts legitimate uses of AI and pushes privacy-conscious users toward workarounds that may be less secure. On the other hand, if companies can demonstrate that AI can work without hoarding personal information, it sets a standard that others may have to follow.

For everyday consumers, the key takeaway is that the privacy battle is not lost — but it depends on the choices we make as users and the tools we support.

What readers can do

You do not need to become a cryptography expert to protect your privacy while using AI. Here are practical steps based on the principles Proton’s CEO outlined:

  • Choose services that prioritize end-to-end encryption. Before signing up for any AI tool, check whether your data is encrypted in transit and at rest — and whether the company can see your inputs. Proton’s own AI features, such as smart compose in Proton Mail, run entirely on the device or through encrypted channels.
  • Limit the amount of personal data you feed into AI models. Do not paste full medical histories, financial documents, or private conversations into generic chatbots unless you are certain about their data handling. Treat AI interactions like conversations in a public space — assume the other party might remember everything.
  • Look for on-device processing. Some AI assistants can perform tasks locally on your phone or computer without sending data to a cloud server. This dramatically reduces the risk of data leaks or reuse.
  • Review privacy policies critically. Many companies have updated their policies to allow training on user inputs. If the policy is vague about data retention or sharing, consider that a red flag.
  • Demand transparency. The CEO’s biggest worry — loss of trust — can only be addressed if companies are open about how their AI models work and what data they use. As a consumer, you can support providers that publish clear, auditable privacy practices.

These steps will not make you immune to every risk, but they significantly reduce the surface area for misuse.

Sources

  • “Privacy in the AI era is possible, says Proton’s CEO, but one thing keeps him up at night” — Spiceworks, June 4, 2026. (Interview with Proton’s CEO.)