The uNHEALTHY
CONTENT FILTER
Our feeds should reflect the body-positive society we've worked hard to build, not reinforce harmful diet culture.
Research shows that watching less than 10 minutes of TikTok content can negatively impact a young woman's body image. Yet despite platform efforts to moderate content, "skinny" and diet-focused posts often slip through the cracks. Using coded language, neutral hashtags and aspirational imagery, this content bypasses existing filters designed to detect nudity, violence or explicit harm.
Introducing the Unhealthy Content Filter — a social media tool designed to do what current Sensitive Content filters don't: identify and hide harmful diet culture and thinspiration content.
92% of UK Gen Z social media users report negative consequences from comparing themselves to others online, and over half experience low-self esteem as a result
- Cybersmile 2023
HOW IT WORKS
1. CONTENT DETECTION
Uses AI to identify posts related to diet culture, thinspiration, and harmful body ideals. Focuses on images, captions, and metadata rather than general content.
2. HASHTAG & LANGUAGE ANALYSIS
Recognises coded language, emojis, and altered spellings (e.g., “#th1nspo”). Flags hashtags and phrases commonly used to promote unhealthy body standards.
3. CONTEXTUAL FILTERING
Differentiates between harmful and neutral content (e.g., fitness motivation vs. extreme
diet promotion). Reduces false positives, so
safe content remains visible.
4. CONTINUOUS LEARNING
AI updates continuously from flagged content and user reports. Adapts to emerging trends or new coded terms in harmful content.
5. USER LEARNING & SUPPORT
Works alongside mental health
guidance and awareness campaigns.