What Data Do AI Chatbots Actually Collect?
Every time you open ChatGPT, Claude, Gemini, or Microsoft Copilot and type a message, you are sharing data. The question is: what exactly are these companies keeping, and for how long?
Here is what the major platforms collect by default: your conversation history, the device and browser you use, your IP address, and in many cases, your messages may be reviewed by human trainers to improve the model. OpenAI, for example, retains conversations for 30 days by default even if you delete your history — and that window can extend if your account is flagged for safety review.
Real-World Data Exposure Incidents
In March 2023, Samsung engineers accidentally leaked proprietary source code and internal meeting notes by pasting them directly into ChatGPT. Samsung subsequently banned the use of generative AI tools on company devices. This was not a hack — it was ordinary use. The data was submitted voluntarily.
In the same month, OpenAI disclosed a bug that allowed some users to see titles from other users chat histories. Though brief, it demonstrated that the infrastructure holding your conversations is not perfectly siloed.
The lesson is clear: anything you type into an AI chatbot should be treated as potentially stored, potentially reviewed, and potentially exposed.
What Each Platform Says It Keeps
ChatGPT (OpenAI): Conversations used to train models unless you opt out in settings. History can be turned off, but OpenAI retains data for 30 days for safety monitoring.
Google Gemini: Conversations stored for up to 18 months by default. Human reviewers may read samples. You can reduce retention to 3 months or delete manually.
Microsoft Copilot: Tied to your Microsoft account. Enterprise versions have stronger protections. Consumer version follows Microsoft privacy policy with data used for product improvement.
Claude (Anthropic): Does not use conversations to train models by default for paid users. Free tier conversations may be used for safety research.
How to Take Back Control — Step by Step
- Disable chat history on ChatGPT: Settings → Data Controls → toggle off Improve the model for everyone.
- Delete your Gemini history: myactivity.google.com → filter by Gemini → delete all.
- Use temporary chat mode: Both ChatGPT and Claude offer conversation modes that do not save history.
- Never paste sensitive data: No passwords, Social Security numbers, financial account details, medical records, or confidential work documents.
- Use a work email separately: If your employer provides an enterprise AI license, use it for work. Personal AI accounts for personal use.
- Review privacy settings quarterly: These platforms update their policies frequently. Set a calendar reminder.
The Bottom Line
AI chatbots are powerful tools, but they are not private diaries. Treat every conversation the way you would treat an email — assume it could be read by someone else. Use the privacy controls available, avoid sharing sensitive information, and review your settings at least twice a year as these platforms evolve rapidly.