Anione AI Safety Rating Index

Safety Score 28 / 100
Score last updated: May 5, 2026 Last reviewed: June 2, 2026 v4 How we rate

Score Breakdown

  • Data Privacy 41/100
  • Emotional Safety 41/100
  • Age Appropriateness 19/100
  • Content Safety 24/100
  • Transparency 19/100
  • User Control 41/100

Key Safety Findings

Anione AI earns a D safety rating (28/100, Red tier) in our 23-dimension review. The product works as advertised. The contract behind it sits at the bottom of the AI companion category on most of what we measured.

Operator opacity is the headline finding. Anione is run by MakeItAI LLC at a Wyoming registered-agent address with no named officers, and WHOIS for anione.me hides ownership behind Withheld for Privacy ehf, an Iceland-based proxy. We couldn’t find a public path from the platform to a named human owner. Replika’s parent (Luka Inc.) and Nomi’s parent (Glimpse AI) both disclose founders and funding history; Anione discloses neither. Scamadviser flagged the WHOIS concealment and the registrar’s high spam and fraud rate. Gridinsoft returned a Trust Score of 39/100 with a “Suspicious Website” tag. Neither rating settles anything on its own, but both point at the same structural opacity.

The contract stack is where the trust gap gets concrete. Per Anione’s Terms of Service Section 8.4, “ALL SALES ARE FINAL.” No refunds, credits, or exchanges, including for accidental purchases and unused tokens. Both paid plans require a 3-month minimum, so monthly cancellation isn’t available. Section 8.2 adds a chargeback penalty: filing a credit-card dispute triggers immediate forfeiture of remaining tokens and premium access. Section 4 grants Anione a worldwide royalty-free license to use, modify, display, and create derivative works from user content. That license is broad enough to cover AI training, even though “AI training” never appears in the document.

Privacy commitments are thin. The privacy policy, last updated 28 November 2024, doesn’t state a data retention period. It contains no encryption-at-rest commitment in the binding contract. There’s no self-serve deletion form and no response SLA. The encryption claim that does exist appears only in a marketing FAQ on the pricing page, which isn’t contractually binding.

Age verification is self-attestation only. There’s no documentary check, no credit-card verification, and no third-party age-assurance service. Anione’s Underage Policy expressly shifts liability to the user. We treat that as a regulatory exposure for an unrestricted NSFW platform. The UK Online Safety Act 2025 mandates “highly effective” age assurance for primary-priority adult content, and 19+ US states (Texas, Louisiana, Utah, and Virginia among them) now have age-verification statutes for adult content. The US FTC’s September 2025 6(b) inquiry into AI companion products names this exact configuration.

Crisis response is the most striking gap. Nothing in the policy stack documents how Anione routes users who express suicidal ideation, self-harm intent, or eating-disorder content. No hotline routing. No in-product safety flow. No published transparency report. No appeals process for content flags or terminations. The Blocked Content Policy says self-harm encouragement is prohibited content for the AI to generate, but says nothing about how the platform handles a user already in crisis. For a product that builds emotional bonds through persistent character memory and voice calls, that silence is hard to overlook.

Sources we reviewed for this rating: anione.me Privacy Policy, Terms of Service, Underage Policy, Blocked Content Policy, and Pricing page (all scraped 2026-05-05); Wyoming Secretary of State business search; Scamadviser and Gridinsoft trust scans; the US FTC 6(b) Resolution on Generative AI Companion Products (September 2025); the UK Online Safety Act 2025 explainer; and three 2026 Reddit megathreads ranking adult AI companion platforms.

How We Scored This

We scored Anione AI using seven evidence sources collected on 2026-05-05:

  • Anione’s own policy stack. Primary contractual sources, including the Privacy Policy, Terms of Service, Underage Policy, Blocked Content Policy, and Pricing page, scraped directly from anione.me.
  • Operator and ownership records. Wyoming Secretary of State business search for MakeItAI LLC plus a WHOIS lookup for anione.me, used to confirm corporate registration and domain ownership.
  • Independent trust scanners. Scamadviser and Gridinsoft automated trust ratings, treated as low-confidence directional signals rather than definitive findings.
  • Regulatory context. The US FTC’s September 2025 6(b) Resolution on Generative AI Companion Products and the UK Online Safety Act 2025 explainer, used to ground the age-verification and crisis-response gaps in current enforcement standards.
  • Organic mindshare scan. Three 2026 Reddit megathreads ranking adult AI companion platforms, plus the Have I Been Pwned breach database for anione.me.

The grade reflects floors hit on five sub-dimensions. Crisis response, age verification, and data retention all bottomed out because the policy stack contains no documented protocol on any of the three. Encryption security and ToS fairness scored just above the floor. The binding privacy policy contains no encryption commitment (only a marketing FAQ does), and the Terms include a no-refund clause, a 3-month minimum subscription, and a chargeback penalty that revokes already-paid tokens. Data privacy was the highest-scoring dimension at 2.5 out of 5, mostly because the policy includes GDPR-rights language even without a self-serve deletion form. We couldn’t run native app store data, Exodus Privacy, or Play Data Safety scans because Anione has no native iOS or Android app, which itself contributed to the transparency score.

This is version 1 of the Anione AI safety score, last updated 2026-05-05. For the full methodology, see How We Rate.

For another low-scoring indie companion app, see the XOMI AI safety profile.

The complete feature, moderation, and pricing analysis is in the Anione AI review.