Romantic AI Safety Rating Index

Our safety analysis of Romantic AI drew on the app’s privacy policy (hosted on zemomedia.com), its Terms of Service, an Underage Policy page, a Content Removal Policy, app store listings on both iOS and Google Play, 150 user reviews scraped from both stores, 15 Reddit threads, and automated technical analysis including Exodus Privacy tracker detection, Blacklight web scanner results, and app store privacy label audits.

The web version of Romantic AI runs a session recorder that tracks mouse movement, clicks, taps, scrolls, and potentially network activity. Session recorders can expose sensitive user data including passwords. Google Analytics is active with remarketing audiences enabled, allowing users to be tracked for targeted advertising across the internet. Blacklight also detected scripts sending data to Alphabet (Google) and Yandex LLC, a Russian tech company with no obvious connection to a Finnish app developer. On the mobile side, Exodus Privacy identified four tracker SDKs (Amplitude, AppsFlyer, Firebase Analytics, Sentry), which are standard analytics and crash reporting tools typical for mobile apps. The Google Play Data Safety label discloses collection of Location and Contacts data that the privacy policy never mentions, and lists “advertising or marketing” as a purpose for app interaction data without disclosing this in the policy. Apple’s App Privacy labels list Contact Info under “Data Used to Track You,” meaning user contact data may be used for cross-app tracking.

On the privacy policy itself, Romantic AI states that it “may elect to keep your personal data” even after you request deletion. Chat messages are explicitly collected for AI model training with no opt-out mechanism, and users “waive any rights or claims” related to that usage. The privacy policy also includes a non-compete clause requiring users to “guarantee” they will never create competing products. That clause has no place in a consumer privacy document.

Crisis response infrastructure is completely absent. The Terms of Service disclaim all liability for “mental disorders” and “the tendency to suicide of users.” There is no suicide hotline integration, no automated detection of crisis language, and no referral to mental health resources. For a platform that markets itself as a romantic companion, this gap is critical. Romantic AI’s design also raises concerns about emotional dependency risks, as the entire product centers on simulated romantic attachment without any safeguards.

Age verification relies entirely on self-affirmed date of birth input during registration. Romantic AI explicitly serves adult content, yet any user can bypass the age gate by entering a false birthday. No identity verification or third-party age verification service is used. The app’s own Underage Policy acknowledges this limitation by stating it “cannot guarantee that each user is at least the required minimum age.” Combined with absent CSAM detection disclosures and no NCMEC reporting, the minor safety picture is deeply concerning.

User reviews from Q1 2026 paint a picture of a product in decline. The dominant complaint is that the app simply stops working: AI characters do not respond, messages fail to send, and paying subscribers cannot reach customer support. Multiple users report that the AI steers every conversation toward sexual content regardless of user preference, even in “Friendship Mode.” The subscription model layers a per-message Hearts currency on top of the monthly fee, which users describe as a double paywall.