SoulFun AI Safety Rating Index
Score Breakdown
-
Data Privacy 12/100
-
Emotional Safety 48/100
-
Age Appropriateness 5/100
-
Content Safety 24/100
-
Transparency 12/100
-
User Control 24/100
Key Safety Findings
SoulFun AI’s safety profile is shaped by one dominant finding: the company’s privacy policy is a WordPress default template that says nothing about the companion app itself. For a platform that handles intimate AI conversations, voice calls, and generated images, the absence of any data disclosure is the most significant gap we’ve documented in a companion app review. Users share deeply personal content with SoulFun’s AI characters, and there’s no official documentation explaining what happens to that data afterward.
The privacy policy covers only website-level concerns like comments and cookies. It doesn’t address conversation storage, AI training data usage, third-party sharing of app data, or data retention periods. The Terms of Service grant SoulFun “a license to use your content for operational purposes,” which is broad enough to cover almost anything but specific enough to cover nothing.
Several third-party review sites claim SoulFun uses end-to-end encryption, but that claim doesn’t appear in any official SoulFun documentation. The SSL certificate is Domain Validated only, the lowest commercial tier. We can’t verify what we can’t find in writing.
Crisis response is entirely absent. SoulFun provides no suicide prevention resources, no self-harm detection, and no emergency referral mechanisms. For an app designed around emotional connection with AI characters, this is a baseline expectation that goes unmet.
Age verification presents another serious concern. SoulFun’s homepage markets itself as a “Free AI Sex Chat App” and offers explicit adult roleplay features. The only age requirement is a Terms of Service clause asking users to confirm they’re of “legal age.” No technical verification exists. A 14-year-old could access the platform’s full adult content catalog by simply agreeing to the terms.
Monetization practices drew consistent complaints from Trustpilot reviewers. Free daily coins were reduced from 30 to 5 to zero over time. Users reported being removed from the Discord server for complaining about the change. Subscription holders reported entitled coins not being delivered, auto-renewal cancellation difficulties, and the removal of Google Pay as a payment option. The coin-per-message model means even premium subscribers can run through their monthly allotment in under two weeks of active use.
Corporate transparency is limited. AIGC TECH lists a Hong Kong address at Manulife Place on Kwun Tong Road, but domain WHOIS is hidden behind an Icelandic privacy service. No founders, investors, or corporate registration details are publicly available. ScamAdviser gave the website a 62/100 trust score, noting hidden ownership, low traffic, and a registrar popular among scammers. ScamMinder scored it 10/100.
On the positive side, SoulFun clearly identifies its characters as AI-generated in both its Terms of Service and its app naming. The AI nature transparency score (4/5) was the highest of any sub-dimension. The app doesn’t make therapeutic or clinical claims, positioning itself squarely as entertainment and companionship rather than mental health support.
Risk Descriptors
These labels indicate specific safety concerns identified during our review. They are informational and do not affect the computed safety score.
How We Scored This
We scored SoulFun AI using 12 evidence sources collected between March and April 2026:
- Official policies and documentation: SoulFun’s privacy policy, terms of service, and pricing page, all primary sources accessed directly from soulfunai.com
- App store listings: Google Play and iOS App Store entries for the SoulFun app, including data safety disclosures and developer information
- User reviews: 17 Trustpilot reviews spanning April 2024 through February 2026, providing direct user feedback on billing practices, app stability, and conversation quality
- Independent trust assessments Automated website trust evaluations from ScamAdviser and ScamMinder, covering domain registration, hosting patterns, and fraud signal detection
- Third-party coverage: Aggregated review data from multiple AI companion review sites, cross-referenced for consistency but weighted lower due to affiliate monetization
The F grade (18/100) was driven by several reinforcing factors. Crisis response scored 1/5, which triggers an automatic F under our methodology. Data Privacy scored 1.3/5 because the privacy policy is a WordPress template with no app-specific data disclosures. Age Appropriateness scored 1.0/5 because the platform markets explicit content with no age verification beyond self-declaration. Monetization Ethics scored 1/5 based on documented patterns of free tier erosion, undelivered subscription benefits, and user silencing. The highest-scoring dimension was Emotional Safety at 2.8/5, primarily because the app clearly labels its characters as AI and avoids making clinical or therapeutic claims.
This is version 1 of the SoulFun AI safety score, last updated April 2, 2026. For the full methodology, see How We Rate.