Share only what the conversation needs: feelings and everyday detail in broad terms can be enough, while identifying and financial details should stay out. When you are sharing personal information with AI, pause long enough to check what the service says about the words you type. If a detail is not needed, leave it out.
Start Broad
A useful AI message can focus on the subject instead of building a profile. You can say how you feel, what sort of day you have had, or that you would like friendly company. You do not need a full name or exact location for those points.
Write at the level you would be comfortable seeing later. “I am at home” may give the context you need without giving an address. “I had a tense day at work” describes the day without naming your employer. The point is not to hide everything. It is to avoid handing over more detail than the conversation uses.
Keep the useful part and cut the label attached to it. Your mood and the kind of company you want may be enough. A workplace, street, account, or person’s name can reveal much more than the conversation needs.
Details to Leave Out
Aiamis's privacy page asks you not to type anything you would not want a third-party service to process. Its examples are passwords, card or bank numbers, identity numbers, your address, and private details about other people.
Keep another person’s private details out too. If a detail could tell a family member, neighbour, or colleague something about you, leave it out when a broader phrase will do. Information about someone else is not yours to share simply because you are the one typing.
Think about why you are adding each detail. If it helps the conversation, keep the broad version. If it merely identifies you or someone else, leave it out. A long prompt is not more valuable than a clear, short one.
Read the Service’s Warning
Google's Gemini Apps Privacy Hub asks users not to enter confidential information they would not want a reviewer to see or Google to use to improve its services, including machine-learning technologies. The warning gives you a plain check: would you be comfortable if a reviewer saw the detail, and do you agree with the service-improvement use it describes?
Replika's privacy policy instructs and strongly encourages users not to share special categories of personal data through the service. The same privacy policy says conversations with an AI companion may lead people to share sensitive information incidentally.
Under that privacy policy, sensitive information included in a message is processed as part of the conversation data, including by third-party AI language model providers, to generate conversational responses and deliver core functionality. An incidental disclosure can therefore still travel through the service’s handling path.
Look Past the Chat Box
The UK National Cyber Security Centre says a query is visible to the organisation providing the model. It warns that queries stored online may be hacked, leaked, or accidentally made publicly accessible. The guidance also notes that an operator may later be acquired by an organisation with a different approach to privacy.
These are risks, not a prediction that any one of them will happen. The practical result is simple: even a casual prompt can contain personal information. Removing details that the conversation does not need gives you more control over what you expose.
What Public Guidance Says
Written for organisations, the Office of the Australian Information Commissioner’s guidance recommends not entering personal information, especially sensitive information, into publicly available AI chatbots and other publicly available generative AI tools because of the significant and complex privacy risks involved.
For your own use, the practical rule is just as direct. If you cannot find a clear explanation of what happens to your words, keep your message general. You do not need to share an exact place, a secret, or someone else’s private life merely to get a useful reply.
Before You Send
Run through a few quick questions:
- Does this detail serve the conversation?
- Is it more precise than you need?
- Does it identify you or another person?
- Do you understand what the service says happens after you send it?
- Can you keep the same point in broader words?
If the answer to a needed question is unclear, take the detail out or do not send it. Reading the service’s own privacy information matters more than trusting a polished reply or assuming that a conversational tone means little is happening in the background.
What Aiamis Does With a Chat
On Aiamis, the privacy page says the messages you send and the replies are stored in the chat service’s database and deleted within 24 hours of the conversation starting. The same page lists a random conversation ID, the time, the page address, the language, and a two-letter country code as stored with the conversation; it says the IP address itself is not stored.
It keeps the conversation in your browser tab until you close the tab or press “New chat.” To produce a reply, your message and earlier messages from the same conversation are sent to an AI model provider, which processes the text on its own servers under its own terms.
Conversations are not forwarded to staff or reviewed as they happen, although the people who maintain the service have technical database access for up to 24 hours. Because conversations are not linked to an identity and are deleted with them, Aiamis usually cannot find a particular conversation to send or delete on request.
One last check is simple: if a detail will not serve the conversation, take it out. Keep the warmth without turning the chat into a personal file.
If you want company from a trained person now, Aiamis's crisis help page lists services, including listening lines for people who are simply lonely. For an AI chat, keep the same plain boundary: share the part that helps, and keep the rest in your hands.