Privacy risks to watch and simple ways to protect yourself

Privacy risks to watch and simple ways to protect yourself

If you’ve asked an AI assistant for help with an email, a recipe, or even sensitive health advice, you’ve probably wondered what happens to that conversation afterward. You are not alone in that concern. In a recent interview with Spiceworks, the CEO of Proton — the company behind ProtonMail and ProtonVPN — said that privacy in the AI era is possible, but one thing keeps him up at night: the lack of transparency and control over how AI companies handle user data.

That sentiment echoes what many privacy researchers have been saying for years. The tools are convenient, but the trade-offs are often unclear.

What happened

On June 4, 2026, Spiceworks published an interview with Proton’s CEO, who acknowledged that while privacy-focused AI is technically feasible, the biggest unresolved problem is user data being collected, stored, and reused without meaningful consent. He pointed to how many popular AI assistants treat every prompt as training material, and how users have little say in what gets retained or shared. The interview did not name specific companies, but the pattern is well known: chatbots, image generators, and voice assistants often send your inputs to servers where they can be repurposed.

The CEO’s main worry was not about encryption or technical limitations, but about the lack of default privacy protections and the difficulty of auditing what companies actually do with data once it leaves your device.

Why it matters

For everyday users, this is not an abstract concern. AI tools are now embedded in browsers, office suites, messaging apps, and search engines. Every time you ask a question or paste a document, you may be handing over sensitive information — your work projects, personal letters, medical details, or private conversations. The terms of service often allow companies to use that data to improve their models, which can lead to your information being re‑exposed to other users or stored indefinitely.

The problem is not that AI is inherently unsafe. It is that the default settings favor the provider, not the user. Without a deliberate effort to choose privacy‑respecting options, you are likely giving away more than you realise.

What readers can do

You do not have to give up AI entirely to protect your privacy. A few concrete steps can reduce your exposure:

1. Choose services with clear, audit‑able privacy policies.
Look for providers that explicitly state they will not train on your data unless you opt in, and that offer end‑to‑end encryption for your conversations. Proton’s own AI services, for example, run on encrypted infrastructure, but other options exist. Read the privacy policy — if it is vague about data retention or sharing with third parties, consider it a red flag.

2. Use on‑device AI whenever possible.
Many modern phones and computers can run AI models locally, meaning your data never leaves your device. Apple’s on‑device processing, open‑source models like Llama 2 that can be run locally, or offline‑capable assistants give you control. For simple tasks like translation or text suggestions, local processing is often sufficient and far more private.

3. Regularly review and delete your history.
Most AI chat services allow you to view and delete your conversation history. Make it a habit to clear it after each session, especially if you discussed anything personal. Also check whether the service allows you to disable training on your data — if it does not, consider switching to one that does. Turn off unnecessary cloud sync features that might keep copies of your inputs.

Conclusion

Proton’s CEO is right: privacy in the AI era is achievable, but it requires active choices from users. The one thing that keeps him up at night — opaque data practices — can be managed if you know what to look for. By picking tools that respect your data, running AI locally when you can, and cleaning up your history regularly, you can enjoy the benefits of AI without handing over more than you intend.

Sources

• Spiceworks, “Privacy in the AI era is possible, says Proton’s CEO, but one thing keeps him up at night,” June 4, 2026.
• Spiceworks, “Enterprise AI initiatives in 2026: Six key dimensions for AI‑ready data,” June 5, 2026.
• Spiceworks, “Nobody knows who regulates your AI yet,” May 20, 2026.
• Spiceworks, “Token shock and the hidden cost of AI consumption,” May 29, 2026.
• Spiceworks, “Your AI vendor can lock you in faster than your cloud provider did,” May 19, 2026.