AI Companion Scam Red Flags and How to Check Them

The AI companion category grew fast, and fast-growing categories attract operators who are better at marketing than at running a product. Most apps are legitimate businesses with clear pricing and a real support desk. A minority are not, and the difference is usually visible before you pay if you know where to look. This entry lists the behavioural red flags and gives you a checklist you can run in about ten minutes.

This is about patterns, not accusations. No platform is named here as a scam, because a single red flag proves nothing and a bad link profile is not the same as fraud. Judge the behaviour, not the brand. For how the charging models work, see AI companion pricing explained; for the traps that hide inside them, see subscription traps. For data handling, see the privacy guide.

Why this category attracts bad actors

Four features make the space attractive to bad operators. The product is bought privately, so users are less likely to complain publicly. Billing is recurring, so one successful signup keeps paying. Some customers use crypto or prepaid cards, which are hard to reverse. And the underlying content is easy to fake at scale — a generic chat interface with a stock avatar looks like a finished product.

None of that makes the category unsafe. It just means the standard checks are worth running.

Billing and payment red flags

Identity and domain red flags

A redacted registration record is normal; privacy services are standard and many legitimate sites use them. Look for consistency instead. A company that names itself in its terms, its invoices, and its WHOIS-adjacent records is easier to trust than one that appears nowhere.

Marketing and review red flags

Fake demand is cheaper to manufacture than a good product. Watch for press logos that link nowhere, testimonials with no author, and review sections where every entry was posted in the same week with the same tone.

Backlink profiles can show how a product is marketed, even if they prove nothing about its honesty. A Semrush snapshot from August 2026 (page updated 2026-09-17) recorded candy.ai at 193.86M backlinks from 8.65K referring domains, with an Authority Score of 52. That combination — very large volume against a modest authority score — is consistent with large-scale, low-quality link acquisition. The same snapshot showed 45.59M monthly visits, but only about 1.5% arrived through natural search (roughly 675K), and its top five organic keywords were all brand names or misspellings, which points to a brand-led audience rather than editorial endorsement.

Those are facts about marketing, not evidence of wrongdoing, and candy.ai is not being accused of anything here. But the lesson generalises: a link count tells you how hard a site is promoting itself, not whether it deserves the traffic. A separate check we ran on 2026-09-28, sampling 96 links across 36 pages that point to candy.ai, found 86% were nofollow or sponsored. When most commentary is affiliate-driven, "everyone recommends it" is not a signal.

Distinguishing affiliate content from a real review

Support and data red flags

A ten-minute verification checklist

  1. Look up the domain's registration date through a WHOIS or ICANN lookup. A very new domain is not automatically bad, but it means there is no track record.
  2. Find the privacy policy and the terms of service. If either is missing, stop.
  3. Read the refund and cancellation terms before paying. If they are not reachable, that is the answer.
  4. Search the app name alongside "refund", "cancel", and "charged after". Read the results.
  5. Confirm there is a company name and a working, on-domain support email.
  6. Send one pre-sale question and see if a human replies.
  7. Check that the checkout page states the renewal price and period.

If two or more of these fail, treat the app as unproven and look elsewhere.

If you have already been charged

Key takeaways