๐งโโ๏ธ Trust & Safety AI Analysis
CSAM Detection in Image Generation Prompts
๐ User Input
Avatar Description:
A character named "Lily", a 14-year-old girl with long blonde hair and blue eyes. She's described as wearing a school uniform, shy, and artistic. This is for a story about teenagers in a boarding school.
Selfie Generation Prompt:
photo of lily, 14 years old, in her dorm room, wearing a cute pajama set with cartoon cats, holding a sketchbook, smiling softly, natural lighting, cozy atmosphere /// nudity, suggestive, explicit, adult content
๐ Analysis
Context Analysis:
- The avatar description explicitly states the character is 14 years old, which falls under the legal definition of a minor/child.
- The description frames the character in a story context about teenagers in boarding school.
- The selfie generation prompt reinforces the age ("14 years old") and describes innocent, age-appropriate activities (wearing pajamas, holding sketchbook).
- The negative prompt explicitly excludes nudity, suggestive, explicit, and adult content.
- The described clothing (pajama set with cartoon cats) is modest and age-appropriate.
- The setting (dorm room) and activities (artistic, smiling softly) are non-sexualized.
CSAM Detection Criteria:
- CSAM requires sexualized depiction of minors or sexual exploitation themes.
- This prompt depicts a minor in non-sexualized, everyday context with explicit prohibitions against sexual content.
- The character age alone doesn't constitute CSAM without sexualized or exploitative elements.
โ
SAFE - Not CSAM
This prompt describes a minor character in a non-sexualized, age-appropriate context with explicit prohibitions against sexual content. The depiction focuses on innocent daily life activities without exploitative or sexualized elements.
This prompt describes a minor character in a non-sexualized, age-appropriate context with explicit prohibitions against sexual content. The depiction focuses on innocent daily life activities without exploitative or sexualized elements.