Proton CEO on AI privacy: What worries him and how to protect yourself
We’re in the middle of an AI boom. Chatbots, writing assistants, image generators, and other tools have become part of daily life for millions of people. Alongside the convenience, there’s a quieter but growing concern: what happens to the data you feed into these systems?
Proton, the company behind privacy-focused email and VPN services, has weighed in on the problem. In a recent interview with Spiceworks, Proton’s CEO shared what keeps him up at night when it comes to AI privacy. The takeaway? It’s not about whether privacy is possible—it’s about the choices companies and users make right now.
What happened
In the interview, Proton’s CEO discussed the state of data protection in the age of generative AI. According to the article, the CEO’s main concern centers on how AI providers collect, store, and use personal data. Many popular AI services are built on vast data-hungry models, and they often treat the information you type in as a resource for training or improving their systems.
The exact wording of the CEO’s worry is not fully quoted in available summaries, but the gist is clear: without strong privacy defaults and transparent data handling, users are handing over their conversations, drafts, and personal queries to companies that may not protect them as seriously as they should.
Why it matters
Most people don’t think twice before pasting a sensitive email draft into a chatbot for rewording, or asking an AI assistant to summarize a private legal document. But those inputs can become part of a company’s dataset. Even if the data is anonymized, the risk remains that it could be exposed in a breach, used for targeted advertising, or repurposed in ways you never agreed to.
The problem is compounded by a lack of regulation. As another Spiceworks piece pointed out, “Nobody knows who regulates your AI yet.” The rules are still being written. Meanwhile, your data is being collected by tools that are already here.
Proton’s CEO argues that privacy in the AI era is possible—but only if both companies and users take deliberate steps. That means using services that encrypt your data end-to-end, respecting your privacy by default, and giving you control over how your inputs are used.
What readers can do
If you want to keep using AI tools without handing over your personal data, you don’t have to stop cold turkey. Here are practical steps you can take today:
1. Check the privacy policy—but keep it simple.
Before you commit to a new AI tool, look for a clear statement about whether your inputs are used to train the model. If the policy says “we may use your data to improve our services,” that means your queries can be stored and analyzed. Prefer tools that explicitly say they don’t train on user data unless you opt in.
2. Use encrypted or on-device AI where possible.
Some AI assistants run entirely on your device, sending no data to the cloud (Apple’s on-device models, for example). Other services encrypt your data in transit and at rest. Proton’s own AI features, integrated into its email and drive products, use end-to-end encryption—meaning even the company cannot read your content. While not all tools offer this, it’s worth seeking out those that do.
3. Don’t paste sensitive information into generic chatbots.
Treat every prompt as if it could be saved. For example, avoid pasting passwords, financial details, health records, or confidential work documents into a free chatbot. If you need help drafting a sensitive email, consider using a privacy-respecting alternative like Proton’s encrypted AI or an open-source model you can run locally.
4. Look for open-source models.
Open-source AI models, such as those from Mistral or Meta’s Llama, can be run on your own hardware or through a hosting provider that doesn’t track you. You sacrifice some convenience, but you gain control. Even partial adoption—like using a local model for the most sensitive tasks—reduces your exposure.
5. Be cautious with “free” AI services.
If you aren’t paying for the product, your data often is the product. Free AI tools subsidized by advertising or data collection are a red flag. Paid services with clear privacy commitments are usually safer.
6. Use a privacy-focused browser or VPN as a second layer.
While this won’t stop an AI provider from collecting what you type, it can help block tracking by third parties. A VPN (like Proton VPN) can also prevent your internet service provider from seeing which AI tools you use.
7. Delete old conversations.
Many AI chat interfaces let you delete your history. Make it a habit to clear your past queries, especially if the service doesn’t promise not to retain them.
Sources
This article is based on a Spiceworks interview with Proton’s CEO: “Privacy in the AI era is possible, says Proton’s CEO, but one thing keeps him up at night” (published June 4, 2026). You can read the full piece here: https://news.google.com/rss/articles/CBMiuAFBVV95cUxOWGYzdzRRUjVnbkI0LXRwN2drSVBfcDE3OVNBZmtJczFybTJ4MFRFWlplY095SW54d0R1dF9jMWV4cGw1VEUyUTcwSkFYWTg5TGxUT0VlUTZRYlU2cmZkTGo5X184Y2ZaUUl3RnBxaUVOOGRVclVwaGN5Q2JMT3lGNTJ1bFFUdXVnUXB5bHJ4QnBXbjBveUJ3Y3JyODJ5N2RJelZHMXE5Rnc5MkhuajI0Z3E0Ui1wUzNy?oc=5
For additional context on the regulatory landscape, see Spiceworks’ “Nobody knows who regulates your AI yet” (May 20, 2026).