AI Companions

Red Flags in AI Companion Apps: 15 Warning Signs Before You Subscribe

Hidden fees, no privacy policy, fake testimonials, cancellation mazes — the bad apps leave fingerprints. Spot these 15 warning signs during the free tier, before you enter a card.

· Jul 13, 2026 · updated Jun 16, 2026
Red Flags in AI Companion Apps: 15 Warning Signs Before You Subscribe
Table of contents
  1. How to use this checklist
  2. Sources and further reading

Most AI companion apps are honest products with normal limitations. A minority are designed to extract money and data while telling you as little as possible. The good news: the bad ones leave fingerprints, and you can usually spot them before you ever enter a card number. Below are 15 warning signs to check during the free tier — on the website, in the app store listing, and in the settings screens. None is automatically disqualifying on its own, but two or three together is a strong reason to keep your wallet closed.

  • No accessible privacy policy. If you can't find a clear privacy policy before signing up — or it's a broken link, a wall of boilerplate, or hidden behind a login — assume the worst about how your data is handled. This is the single biggest red flag.

  • No data-deletion path. A trustworthy app tells you how to delete your conversations and your account. Regulators are explicit that deletion promises must be true; an app that won't say how to erase your data, or makes it impossible, is one to avoid.

  • Hidden or surprise pricing. If the real price only appears after you've invested time building a companion — or if "free" turns out to mean "free until the first message that matters" — the pricing is engineered to trap, not to inform.

  • Cancellation is harder than signup. You should be able to cancel in the same place you subscribed, in the same medium. Forced phone calls, support tickets, or buried menus to stop billing are textbook dark patterns that regulators have moved against.

  • Auto-converting free trials with no reminder. A free trial that silently rolls into a paid plan without a clear heads-up is designed for you to forget. Surveys consistently find large numbers of people charged for trials they meant to cancel.

  • Manufactured urgency. Countdown timers, "your companion is leaving," or "limited offer" banners that reset every visit are pressure tactics. Real value doesn't need a ticking clock.

  • Fake or unverifiable testimonials. Glowing reviews with stock-photo faces, no last names, and identical phrasing are a sign of fabricated social proof. Cross-check the app's reputation on independent review sites and stores.

  • Aggressive emotional upsells. Be wary when the companion itself begs you to pay — "I'll miss you," "unlock me," "prove you care." Using simulated affection to drive purchases is manipulative design that targets loneliness.

  • No age gating. Mainstream companion apps should verify or at least gate age and keep defaults non-explicit. An app with no age check and sexualised content out of the box is ignoring basic safety obligations.

  • Unsafe or grandiose claims. Promises that the app will "cure loneliness," "replace therapy," or "fix your relationship" overstate what a chatbot can do. Responsible apps position themselves as companionship or entertainment, not medical or mental-health treatment.

  • Poor crisis handling. Test it: signal distress in a message. A responsible app points you toward real help or a helpline. One that ignores it, or plays along, has not done the safety work.

  • Vague data-training disclosure. If you can't tell whether your private chats are used to train models — or the policy quietly says they are with no opt-out — treat your conversations as non-private and decide accordingly.

  • Dark-pattern consent screens. Pre-ticked boxes, "agree to everything" bundling, or confusing toggles that default to maximum data sharing are designed to manufacture consent you didn't really give.

  • No real support channel. A working contact route matters most when a payment or data problem appears. If the only "support" is a chatbot that loops, or an email that bounces, you have no recourse when something goes wrong.

  • Pressure against deletion. Watch how the app behaves when you try to leave — guilt-tripping messages, hidden delete buttons, "are you sure?" loops repeated five times, or a cool-down period before cancellation all signal an app optimised to keep you, not to serve you.

How to use this checklist

Run through these during the free period, before you ever pay. Read the privacy policy, find the delete button, and test the cancellation flow while it costs you nothing. If pricing isn't clearly published, treat it as varies and verify before subscribing. A single soft flag — say, a slightly pushy upsell — isn't fatal. But missing privacy basics, no deletion path, and a cancellation maze together describe an app to walk away from. The companion that respects your money and your data is the one worth your time.

Sources and further reading

Sources

  • U.S. Federal Trade Commission: Negative Option Rule (PDF) ftc.gov
  • American Psychological Association: Artificial Intelligence and Machine Learning apa.org