starryai Enterprise Risk Assessment | AI Tool Risk

Back to AI Tools

starryai

Risk Overview

Moderate Risk

3.9

/10

Image Generation

Top Risks

Face data (biometric) used to 'improve image generation models' — this is a form of training use of biometric data; third-party analytics via AppsFlyer and Facebook SDK; no publicly confirmed security certifications; full ToS difficult to access (hosted externally).

Mitigations

Avoid using face-related features if biometric training is a concern; review and accept ToS via the linked getterms.io page; understand that non-face generated images appear safe from training use; strong IP ownership makes commercial use safe.

Get started

Is your team using

starryai

safely?

Book a Demo Free Assessment

Highest Risk Categories

The highest-scoring risk categories identified in this assessment.

3rd-Party Sharing

Does not sell personal data. Shares with service providers (AWS, Cloudflare, OpenAI, Replicate, Mixpanel, AppsFlyer, Intercom, Stripe) contractually restricted. AppsFlyer used for marketing attribution which involves device identifiers. Facebook SDK used for analytics. No third-party ad sales.

/10

Security Posture

No public SOC 2 or ISO 27001 certification found. Uses standard cloud security (AWS, Cloudflare). Face data encrypted and access-restricted. No known major public breaches. No explicit bug bounty program mentioned. Basic security posture documented in privacy policy but no independent audit certification.

/10

Training Use

Starryai explicitly states user-generated content may be used 'to improve the accuracy and quality of our image generation models' — face data collected via face features is used 'to improve the accuracy and quality of our image generation models.' This is opt-in when using face features, but default for general in-app activity. OpenAI API data (used as backend) not trained per OpenAI policy.

/10