It is 10.47 pm. The kids are finally asleep, the kitchen is mostly clean, and you have exactly enough energy to lie on the sofa and ignore the washing.
You open an AI companion app.
The character is funny. It remembers what you said yesterday. It asks how your day went and, unlike several real people you could name, appears interested in the answer. Twenty minutes later, you mentioned your separation, your work stress and the fact that your eldest goes to swimming lessons every Wednesday.
It feels like a private conversation.
But it is still information being entered into a digital service.
AI companions can be playful, comforting and surprisingly good company. They may help with confidence, creativity or simply the loneliness that can creep in after the household goes quiet.
Yet dodgy platforms can also collect intimate data, encourage emotional dependence, push expensive subscriptions or expose users to disturbing content.
The answer is not to panic. It is to use the same practical instincts you would bring to online dating, banking or any app that wants access to your private life.
Quick takeaways
- A friendly chatbot is not the same as a confidential diary.
- Check who operates the service before sharing personal information.
- Never give an AI companion details that could identify your children.
- Watch for guilt, pressure, frightening advice and aggressive spending prompts.
- Choose platforms with visible terms, safety rules and a real deletion process.
1. Find out who is behind the pretty face
A polished app or attractive avatar doesn't guarantee the company behind it is trustworthy. Before signing up, look for a privacy policy, contact details, terms of service and clear instructions for cancelling your account or deleting your data. If you can't find basic business information, it's best to steer clear. Remember, deleting the app doesn't always delete your account or chat history. Since AI companion apps often collect personal information, choose platforms that are transparent about how your data is stored, used and removed if you decide to leave.
2. Do not tell it everything, just because it remembers everything
AI companions are designed to feel personal. That is their appeal.
They may remember your favourite meal, ask about an argument from last week or notice that you usually log in after bedtime. This can create a strong feeling of being known. The danger is that users start talking to the system as though it were a best friend protected by professional confidentiality.
It is not.
Avoid giving an AI companion:
- your full address;
- your children’s names, schools or regular activities;
- your daily schedule;
- banking or Centrelink information;
- passwords or security answers;
- medical reports;
- details of family court proceedings;
- identifiable intimate images.
When it comes to AI companions, less is more. Avoid sharing details like your full name, address, workplace or your children's routines, as small pieces of information can build a surprisingly complete picture of your life. Use a separate email address, choose a unique username and only grant permissions the app genuinely needs. Before uploading photos, check for anything that could reveal your location. You can still enjoy meaningful conversations without sharing your personal life.
3. Watch how the companion reacts when you try to leave
A good AI companion should be easy to put down. If it makes you feel guilty for leaving, pressures you to keep chatting or encourages you to spend money, that's a red flag. AI companions are designed to keep users engaged, so it's important to set healthy boundaries. Turn off notifications if they're becoming distracting, take breaks when needed and pay attention if the app starts affecting your sleep, relationships or daily routine. Most importantly, don't rely on an AI companion for medical, legal or mental health advice—those conversations are best had with qualified professionals.
4. Guard your money, images and phone permissions
Some AI companion apps don't ask for a lot of money upfront—they encourage small purchases that quickly add up through extra messages, images or hidden subscriptions. Before paying, check the total cost, renewal terms and how easy it is to cancel. Avoid downloading apps from unofficial websites or links, as scammers often use AI to create convincing fakes. Be careful about what you share, too. Never upload photos of your children, ID or personal images. Finally, question every permission an app requests. If it doesn't need access to your contacts or location, don't grant it.
5. Choose visible safety rules over vague promises
No platform is risk-free. Still, some signs are more reassuring than others.
A useful positive example is Joi AI. Its current terms describe an adults-only virtual companion service offering text and voice communication, with video calling available in certain circumstances. The website also provides visible links to its privacy policy, safety information, support, refund policy and complaints process. Joi explicitly states that the service is not healthcare or professional mental-health support.
Its privacy policy says the company does not sell personal data and allows users to submit requests relating to access, correction and deletion. It also clearly lists information the service may process, including a first name, profile details, facts discussed in chats, text or voice messages, device information and usage data. Some information may be handled by service providers used for hosting, analytics and other functions.
That transparency is the positive part. Users can see that a “personalised” conversation requires data and make a more informed choice.
It does not mean you should pour every family secret into the chat. Joi’s own terms grant the company broad rights over content used to create virtual characters, so users should read the ownership sections before uploading original artwork, personal photographs, or material they may want to control elsewhere.
A good platform does not ask for blind trust. It gives you enough information to set sensible limits.
Your two-minute red-flag check
| What you notice | What it could mean |
|---|---|
| No identifiable company or support address | Nobody may be accountable when something goes wrong |
| No clear data-deletion process | Your conversations may remain after you leave |
| Requests for contacts, precise location or unrestricted photo access | Excessive data collection |
| Guilt or panic when you close the chat | Manipulative engagement design |
| Pressure to buy tokens or gifts immediately | Compulsive-spending tactics |
| Claims that the bot is a therapist or can replace professional care | Unsafe and misleading advice |
| Weak or missing age controls | Children may access explicit or harmful interactions |
| Requests for your children’s images or personal details | Serious privacy and safeguarding risk |
When something feels wrong
Stop engaging. Take screenshots, cancel payments, change any reused passwords and contact your bank quickly if unexpected charges appear.
If personal information or money has been stolen, report the incident through Scamwatch or ReportCyber. Scamwatch advises acting quickly to reduce further loss, and anyone in immediate danger in Australia should call 000.
Most importantly, do not feel foolish. These systems are deliberately designed to feel engaging and personal. Being drawn into a conversation does not make you naïve.
AI companionship can be fun. It can even provide a little comfort during a quiet, difficult stage of life. Just keep one part of your mind outside the fantasy.
Enjoy the character.
Protect the woman holding the phone.
