Privacy in the AI Era: What Keeps Proton’s CEO Awake—and How You Can Protect Yourself
In a recent interview with Spiceworks, Proton’s CEO Andy Yen offered a refreshingly frank take on privacy in the age of AI. On one hand, he believes that meaningful privacy is still achievable—even as AI tools become embedded in our daily workflows. On the other, he pointed to a specific threat that keeps him up at night: the way AI providers are quietly shifting from being tools to being data brokers.
The concern is not about the technology itself, but about the business models behind many popular AI services. When you paste a confidential email into a chat interface, or ask an AI to summarise a private document, that data often gets stored, analysed, and potentially reused to train future models. The default in many cases is that your data is not your own.
What specifically worries Proton’s CEO
Yen highlighted what he sees as the most dangerous trend: the lack of transparency around how user data flows through AI systems. Many consumers assume that because they are logged into a service, their conversations are private. In reality, providers may share inputs with third-party APIs, retain logs indefinitely, or use machine learning to extract patterns from personal information.
The one thing that keeps him up at night, according to the interview, is the combination of user ignorance and platform opacity. Users often do not know what they are giving up, and companies are not required to explain it clearly. This is not a hypothetical risk—it is already happening.
Why this matters for everyday users
The real-world impact is tangible. Imagine using a free AI assistant to draft a sensitive email about a medical appointment or a financial negotiation. That text could end up in a training set, searchable by others, or sold to advertisers. Even when companies claim to anonymise data, research has repeatedly shown that anonymisation is often reversible.
The problem is compounded by the fact that many AI tools are integrated into other services you already use—email, calendars, notes—making it difficult to know where your data ends. The convenience is real, but so is the exposure.
Practical steps you can take today
You do not need to abandon AI entirely to protect your privacy. Here are concrete actions that privacy-conscious users can take, many of which align with Proton’s own approach:
Choose AI services that use end‑to‑end encryption. Proton itself offers Proton AI, which runs locally on device or uses encrypted processing so that even Proton cannot see your prompts. Look for tools that explicitly state they do not store or train on your data.
Read the privacy policy—specifically the “data use” section. Many services bury important details. If the policy says they may “share with affiliates” or “use for service improvement,” assume your data is not private. If you cannot find a clear statement, contact the provider.
Use local AI when possible. Running models locally (e.g., via tools like Ollama, LM Studio, or Proton VPN’s built‑in local AI) eliminates the need to send your data to any server. It may be slower, but it is the most private option.
Avoid pasting sensitive content into general‑purpose chatbots. Even if the chatbot claims to be private, treat it as a public forum until proven otherwise. Create a separate, isolated account for personal use if needed.
Adjust settings to opt out of training. Some services now offer a toggle to prevent your conversations from being used for training. Enable it, and check it regularly as defaults can change.
How Proton and others are addressing these concerns
Proton’s response has been to build AI features that work entirely on the user’s device or within its encrypted ecosystem. Their Proton AI for business and personal use is designed to never see your plaintext. Other companies like Apple (with on‑device processing) and some open‑source projects offer similar guarantees, though each has trade‑offs.
The key takeaway from Yen’s interview is that the problem is not unsolvable. It just requires users to be informed and companies to be transparent. Until regulation catches up, your best defence is to choose tools that put privacy first by design.
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’s official privacy principles and product documentation.
- Independent research on data retention in AI services (various academic sources).