What never to paste into an AI chatbot
Treat a chat like a message to a company that may store it and, depending on your settings, have people review it or use it to improve its models. Google’s Gemini privacy notice says this directly: don’t enter confidential information you wouldn’t want a reviewer to see.
- Passwords, sign-in codes, recovery codes and security answers.
- Card numbers, bank details and government ID numbers.
- Medical, legal or financial records with names attached.
- Other people’s private information, photos or messages without their permission.
- Confidential work or school material your organisation hasn’t approved for AI tools.
When you need help with something personal, swap in made-up names and numbers. The AI usually needs the shape of the problem, not your real details.
Privacy settings worth checking
Settings change often, so check your app’s own help pages. At the time of writing:
- ChatGPT: Settings → Data Controls → “Improve the model for everyone” controls whether your chats help train models. Temporary Chat keeps a conversation out of your history and memory; OpenAI says temporary chats aren’t used for training and are deleted after 30 days.
- Gemini: the “Keep Activity” setting controls whether your Gemini activity is saved and used to improve Google’s services. Google says human reviewers read some conversations.
- Claude: the Model Improvement privacy setting controls whether your chats are used to improve Claude. Anthropic says Incognito chats aren’t used for training.
- Memory: ChatGPT and other assistants can remember details between chats. Review what they remember, and turn memory off if you don’t want it.
Check what AI tells you
AI chatbots can state wrong things with total confidence, invent sources and links, and agree with you just because you asked a leading question. ChatGPT shows its own reminder that it can make mistakes.
- Ask for sources, then open them. Make sure the source exists and actually says what the AI claims.
- Check important facts somewhere else: an official site, a trusted publication or a professional.
- Watch for dates. An AI’s knowledge can be out of date unless it is searching the web, and even then it can misread pages.
- Ask the same question in a neutral way. “Is this a good idea?” gets a more honest answer than “Why is this a great idea?”
- Let a person decide anything about health, money, legal matters or safety. Use AI to prepare questions, not to replace a doctor, lawyer or advisor.
Safe habits for kids and teens
Check each app’s age rules, talk about what not to share, and remind young people that a chatbot can sound caring without being a person. AI companions shouldn’t replace trusted adults or friends.
How Mekat helps
Mekat teaches these habits in short lessons with practice: what never to paste, chat history, memory and data settings, checking AI citations and links, and hidden instructions and account safety. See the full learning path.
Quick answers
Is it safe to use ChatGPT?
Yes, for most everyday tasks, if you don’t share sensitive personal information and you check important answers. Review the data controls in settings and turn off model training if you prefer.
Does ChatGPT use my conversations for training?
It can, depending on your settings. In ChatGPT, turn off “Improve the model for everyone” in Settings → Data Controls, or use Temporary Chat, which OpenAI says isn’t used for training.
Can I trust the sources ChatGPT gives me?
Not without checking. AI chatbots can cite sources that don’t exist or don’t say what they claim. Open each source and confirm it before you rely on it.
Sources
Written by the Mekat team at Martinez Solutions LLC, Glendale, Arizona. Spotted something out of date? support@martinezsolutionsaz.com