SimSimi Safety Rating Index
Score Breakdown
-
Data Privacy 12/100
-
Emotional Safety 50/100
-
Age Appropriateness 19/100
-
Content Safety 22/100
-
Transparency 29/100
-
User Control 36/100
Key Safety Findings
We scored SimSimi from 17 distinct evidence sources gathered between April 28 and April 30, 2026. Primary documents drawn from SimSimi Inc.’s public-facing materials: the canonical English consumer Terms and Conditions at blog.simsimi.com, the o.simsimi.com privacy policy, the workshop.simsimi.com developer documentation, and the bad-words policy page (last updated 2017). The iOS App Store listing by SimSimi Inc. and the Google Play listing (com.ismaker.android.simsimi) provided current age ratings, store-disclosed app privacy data, version history, and the developer’s public-facing positioning. We pulled review samples from both stores spanning the 12-month window, surfacing a bimodal sentiment distribution (40% 5-star plus 43% 1-star) with a thin middle and a recent decline in average ratings.
Three independent technical scans corroborated the privacy and tracking surface. Exodus Privacy detected 12 advertising and analytics SDKs on the Android app version analyzed, including AppLovin, Mintegral, Pangle, Taboola, MixPanel, and ironSource. Blacklight’s scan of the simsimi.com marketing website returned 0 ad trackers and 0 third-party cookies, a strong contrast with the Android app’s tracking surface. Have I Been Pwned queries returned zero documented breaches at simsimi.com or related domains.
The historical record was the dominant evidence source for content-safety scoring. Three separate national disruptions are documented: a 2012 Thailand political controversy with brief US App Store removal, a 2017 Ireland voluntary suspension following coordinated school, police, and media warnings (BBC, Irish Times, PSNI, Webwise.ie), and a 2018 Brazil voluntary suspension after an Estadao investigation and a SaferNet alert. The 2025 Graphika report on minor-sexualization chatbot platforms did not flag SimSimi specifically.
Recent product changes were captured from public release notes. Terms of Service Section 2.3 added an explicit AI disclosure in December 2025; Section 2.4 added clarified reporting flows including AI-generated content in January 2026. iOS release 9.1.6 in January 2026 introduced region-specific age restriction language in the public release notes. The bad-words policy page itself has not been substantively updated since 2017. Cross-references between the privacy policy, the Play Store Data Safety declaration, and the Exodus tracker scan surfaced inconsistencies on location collection and the count of named advertising partners, which we held the score back for in the transparency dimension.
How We Scored This
We scored SimSimi using 17 evidence sources collected between April 28 and April 30, 2026:
- SimSimi Inc. primary documents: privacy policy, canonical English consumer Terms and Conditions, developer workshop terms, bad-words policy
- App store listings on iOS App Store (SimSimi Inc.) and Google Play (com.ismaker.android.simsimi), plus 12-month review samples from both stores
- Independent technical scans: Exodus Privacy tracker analysis, Blacklight website scan, Have I Been Pwned breach queries
- Historical record: BBC, Irish Times, PSNI, Webwise.ie coverage of the 2017 Ireland suspension; Estadao and SaferNet coverage of the 2018 Brazil suspension; archived 2012 Thailand reporting
- Recent product changes: ToS Section 2.3 (December 2025 AI disclosure), Section 2.4 (January 2026 reporting flow), iOS release 9.1.6 release notes
- Comparative reference: 2025 Graphika report on minor-sexualization chatbot platforms (SimSimi was not named)
The score reflects structural concerns about the crowd-taught moderation architecture. The historical record across Thailand, Ireland, and Brazil shows the architecture has produced repeat youth-safety harms across cultures, not isolated incidents. We weighted those bans heavily in minor content moderation, minor safeguards, and safety reporting sub-dimensions. The bad-words policy itself has not been substantively updated since 2017, which scored down on transparency and active moderation.
Recent improvements moderated some scores but could not overcome the structural signals. The December 2025 AI disclosure and January 2026 reporting flow were credited in AI nature transparency and reporting dimensions. The clean website surface (zero ad trackers on simsimi.com) was credited as evidence of intentional separation between the marketing site and the heavily-trackered Android app, but the app itself carries 12 advertising SDKs that pulled the data minimization and third-party sharing dimensions down. Privacy Shield references in the privacy policy (invalidated by Schrems II in 2020) and discrepancies between the privacy policy and the Play Store Data Safety declaration on location collection scored down on transparency.
Score Engine v3.1, scored 2026-04-30. See How We Rate for the full 23-dimension framework.
Version History
Initial AI scoring from evidence — 2/3 panel (GPT-5.4 + MiniMax M2.7); Gemini failed. Auto-F triggered by crisis_response=1.