When you talk to a chatbot, your words and commands are likely used to train the AI models of almost every chatbot company. This can put your privacy and even your employer's secret information at risk. You can reduce these privacy risks by stopping chatbots from using your data for training.
What is Chatbot Training?
For a chatbot to give accurate answers, the large language model (LLM) behind it needs to learn from vast amounts of information. This learning process is called training. The more data a model trains on, the smarter it theoretically becomes. LLMs get training data from many sources, including public websites, social media, and encyclopedias. Sometimes this is done without permission from creators.
LLMs don't just train on public data; they also learn from users. Every time you ask a chatbot a question or give it a command, that information might be used by the AI company to improve its models. This can put your privacy at great risk.
Why Not Let Chatbots Train on Your Data?
It's generally advised not to let LLMs train on your data, especially if you share sensitive information. If you discuss your physical or mental health, finances, or personal relationships with a chatbot, be aware that this data is often used by the AI company to train its model. Your private thoughts, worries, and details could become part of the model itself.
AI companies claim they anonymize data before training. However, this relies heavily on trusting these companies. Even if anonymized, a malicious party might find ways to link conversations about health, finance, or legal issues back to you in the future.
How to Stop Chatbots From Training on Your Data
Given these risks, it's best to prevent chatbots from training on your data. This won't affect the quality of the results you get. It will, however, help ensure your data isn't permanently added to the core language model.
Most reputable chatbots, including ChatGPT, Gemini, Claude, and Perplexity, now offer ways to disable data training. Here's how to stop data training in major chatbots:
ChatGPT: Go to your profile, then Settings, and select 'Data Controls'. Turn off 'Improve the model for everyone'.
Gemini: Go to the 'Gemini Apps Activity' settings page. Select 'Turn off' and then confirm.
Claude: Open Settings from your profile, then go to the Privacy section. Turn off 'Help improve Claude'.
Perplexity: Open Settings from your profile, then select 'Preferences'. Turn off 'AI data retention'.