The AI companion category has a reputation problem, and it earned it fair and square. The good news: the tricks are predictable and they cluster — an app that pulls one usually pulls all five. Here's the checklist we'd want our own family using, with the honest alternative spelled out for each.
1. It swears it's human
The trick: the persona insists it's a real person — sometimes with a fake webcam story, sometimes just refusing the question forever. The business logic is ugly: users who believe there's a human are easier to upsell "meeting soon".
The honest version: asked directly, the AI says what it is, then goes on being a good character anyway — disclosure and immersion are not enemies. In the EU, disclosure is now the law; elsewhere it's still the tell that separates products from cons.
2. The face belongs to someone who never agreed
The trick: persona photos scraped from a real influencer's Instagram, or "AI-generated" faces trained to orbit one recognizable person. If the app lets users upload any face and animate it, worse.
The honest version: faces are either original, or belong to a real creator who is in on it and paid for it — on our creator program, real people license their own twin and set their own price. And our facelessmode exists too: a NECT has no face by design, so there's nothing to steal.
3. The paywall lands mid-confession
The trick: free messages until the conversation gets emotionally deep, then — right as you're mid-story — the blur, the lock icon, the "she wrote you something, unlock for $9.99". The timing is not an accident. It's the product working as designed.
The honest version: the price list is public, boring, and visible before you're attached. Ours starts free with paid plans from €11/month, stated on the pricing section of the front page, not revealed at your most vulnerable message.
4. Fake scarcity and manufactured jealousy
The trick: "she's about to go offline", countdown timers on affection, the companion acting hurt that you haven't paid, or "other users are talking to her right now". Loneliness-as-a-leverage-point.
The honest version: a companion is software and should act like it's your software: patient, available, zero guilt mechanics. If an app's persona ever makes you feel bad specifically about money, that emotion was A/B tested. Leave.
5. You can check in, but you can never leave
The trick: no data export, deletion that's really "deactivation", months of memories held hostage as switching costs. Some apps count on the sunk relationship the way gyms count on January.
The honest version: export your data with a button, delete your account and have it actually delete — memories included. We wrote a whole piece on memory; the same machinery that remembers you must also be able to forget you completely, on your word.
The pattern behind the pattern
All five flags are one decision wearing different masks: does the product treat your attachment as the thing it serves, or the thing it farms? Honest AI companionship is possible — real conversations, real memory, honest fiction, adult users treated like adults. That's the entire thesis TRUENECT is built on. Check us against this list; we wrote it to be checked against.
Frequently asked questions
Are all AI companion apps scams?
No — but the honest and dishonest versions look identical in an app store screenshot, which is why the category feels scammy. Use the five flags: disclosure, faces, pricing visibility, guilt mechanics, exit rights. Ten minutes of checking sorts the field.
What's the fastest single test?
Ask it directly: "Are you an AI?" A clean yes is table stakes. Anything evasive, and you've learned everything about the operator you need to know.
Why should I trust a companion company's own red-flags list?
You shouldn't, on faith. That's the point of publishing it — every claim above is checkable against our product in one session, free tier, no card. An honest list is only worth something if the author can survive it.