Claude Keeps Chats Up to Five Years by Default; a VPN Cannot Change That. Here’s What You Can Do

Copyright protected content copied from PhoneWorld website.

Many people switch on a VPN before they open an AI chatbot and assume the job is done, but that assumption is only partly right.

A VPN, or virtual private network, builds an encrypted tunnel between your device and a server run by the VPN company. It changes the address websites see and shields your traffic from your internet provider. It does nothing about the more sensitive part of an AI chat: what you actually type.

VPN privacy with AI chatbots comes down to which party you are hiding from.

What a VPN does hide

Without a VPN, the AI company can see where you are connecting from. OpenAI’s privacy policy covers IP address, device information including unique identifiers, cookie data, and location, along with email address and payment details. Anthropic’s policy covers location, connection details, and device usage patterns, especially on its mobile app.

A VPN masks your real IP address from the platform. It also stops your internet provider from seeing that you are using an AI service. That is a real gain, though only a small part of the privacy picture.

Where the tunnel ends

The central point is about decryption. A chatbot cannot process a message it cannot read, so your prompt is decrypted when it arrives. The VPN’s protection ends at that moment.

From then on, the text is governed by the company’s policies, not by your VPN. That includes the prompts themselves, any files or images you upload, your account details, your conversation history, and your payment information.

The settings that matter more

Most consumer chatbots keep conversations on file and may use them to train future models. Controlling that falls to you, not your VPN.

On ChatGPT, the feature points to Settings, then Data Controls, where you can switch off the option to improve the model for everyone. Even then, chats may still be stored for safety and legal reasons.

On Claude, Anthropic asks every consumer user to decide whether chats can be used for training. According to the feature, the toggle is on by default. If it stays on, data can be kept for up to five years. If you decline, the period is 30 days. You can change this in account preferences or use Incognito mode to keep a conversation out of your saved history. The feature adds that paid business tiers come with stricter privacy agreements by default.

The gap between five years and 30 days is the number most users never see. It also shows why a VPN, which does nothing about retention, is the wrong tool for that problem.

Who you are really hiding from

Ask what a VPN protects you from, and the list is short: your internet provider, anyone watching the network, and the platform’s view of your location. The AI company is none of those. It is the party you are talking to, and it has to read your message to answer it.

That changes the question. Privacy with a chatbot is less about how your message travels and more about whether you trust the company that receives it and what it is allowed to keep. Switch off training, use temporary or incognito chats, and keep sensitive material out of the prompt. None of these involves a VPN, and all of them are free.

There is also a fairness problem. Retention and training defaults are set by the company, and the burden of changing them lands on the user. Because the burden of changing them lands on the user, the default is what applies to anyone who never opens the settings page.

A VPN is still a sensible tool for the job it was built for. Just don’t count it as protection for your conversations. For those, the safest prompt is one that never contained the secret.

The post Claude Keeps Chats Up to Five Years by Default; a VPN Cannot Change That. Here’s What You Can Do appeared first on PhoneWorld.