Major AI chatbots collect your data by default—here's how to opt out and protect your privacy

AI Chatbots Are Collecting More Data Than You Think

AI data collection has expanded significantly as companies race to improve their models. By default, most AI chatbots—including Google's Gemini, OpenAI's ChatGPT, and Anthropic's Claude—store your conversations and use them to train large language models 1

1

. Google quietly updated its search privacy settings last month, automatically opting all customers into expanded AI training that now includes images, voice searches, and videos uploaded to Google's search services 2

2

. This shift raises significant privacy concerns, particularly for users who share sensitive information with AI chatbots for therapy, health advice, or workplace discussions.

The data collection practices extend beyond simple text conversations. Google's updated privacy policy states that saved media "may be used to develop and improve Google's AI models and technologies, as well as the Google services that use them" 2

2

. This includes screenshots from Circle to Search, photos, and voice recordings from Search Live or Google Translate 2

2

. If your data has already been used to train an AI model, there's no undo option, making preemptive action critical 1

1

.

How to Opt Out of AI Chatbots Collecting Your Data

To protect your privacy, you need to disable these settings across multiple platforms. For Google and Gemini, navigate to myactivity.google.com, click Web & App Activity, and turn it off 1

1

. Additionally, open the Google app, access Search Services History, and uncheck "Save media" 2

2

. Within Gemini itself, log in, click the Settings cog, navigate to Activity, and set it to Off 1

1

.

For ChatGPT users, log in to your account, click your profile icon, select Settings, then Data controls, and switch off "improve the model for everyone" 1

1

. Claude users should click their profile icon, go to Settings, select Privacy, and toggle "Help improve our AI models" to off 1

1

. These changes only affect future AI training—if you've previously allowed data collection, your information will still be used in training that's already in progress 1

1

.

Delete Existing Conversation Data to Limit Exposure

Opting out alone isn't enough. You must also delete existing conversation data that companies have already stored. In Gemini, access the Gemini Apps Activity page and click the X next to individual chats or use the Delete drop-down to remove conversations from the last hour, day, all time, or a custom range 3

3

. You can also set up an auto-delete option for 3 months, 18 months, or 36 months 3

3

.

For Claude, click Chats on the left sidebar, select all conversations, and delete them 1

1

. ChatGPT offers a dedicated deletion button within the Data controls menu 1

1

. Both Anthropic and OpenAI store deleted chats for up to 30 days in their back-end systems before permanent removal 1

1

. According to Anthropic's privacy policy, new and existing chats won't be used to train models going forward once you disable the setting, but your data will still be used in training currently in progress 1

1

.

Privacy Concerns Mount as AI Training Expands

The implications of unchecked AI data collection extend beyond personal discomfort. Consider a scenario where you discuss leaving your job with a chatbot, sharing your company name and boss's identity. That chat history trains the model, and days later, your boss asks the same AI about employee loyalty—potentially exposing your plans 2

2

. Voice recordings present another risk: your voice could be used to train models that later generate audio clips of you saying things you never said 2

2

.

Google's Gemini collects prompts, shared files, videos, photos, browser pages, transcripts, Gemini Live recordings, and custom instructions from your Gems 3

3

. Even if you don't use the chatbot directly, you're likely interacting with the AI model through Google's base features 3

3

. The only way to prevent Gemini from training on your chats is to turn off Google's history retention entirely, which means you won't be able to view or refer back to prior conversations 3

3

.

What to Watch as Data Collection Practices Evolve

As AI companies continue updating their privacy settings, users should monitor their accounts regularly. Google is training AI on your data by default, and other companies are following similar paths. Watch for email notifications about privacy policy changes and review your privacy settings quarterly. Personal data shared with AI chatbots today could influence model behavior for years, affecting not just your experience but potentially exposing information to others. The balance between AI convenience and privacy protection requires active management—companies won't prioritize your privacy unless you take action.