[Verse 1] Swipe up, swipe up, infinite feed For You page learns your speed Watch time measured to the frame Replays counted, pause length saved Like, share, follow, linger rate Completion ratio calibrates Device model, IP band Keyboard language, time of scan Self-reported birth year field Age gate asks, but signals build Behavioral inference layered on Confidence score moving strong [Pre-Chorus] Interest clusters spin in space Vector embeddings track your pace Teen cohort segmentation lines Derived from scroll and screen time [Chorus] TikTok teen signals Watch-time weighted score Micro-gestures into features Fed back into the core Algorithmic profiling Minor risk and trend Behavioral clustering Looped without an end TikTok teen signals Recommendation stack Personalized ranking model Optimizing you back [Verse 2] Two thousand nineteen settlement filed FTC and Musical.ly child COPPA violations cited there Civil penalty, compliance repair Now regulators in the EU Under DSA review Transparency reports required Systemic risk inquiry pursued State attorneys general ask About design and mental health Engagement incentives tuned Toward retention over welfare Age estimation models train On patterns not just forms Facial geometry, engagement bands Behavioral age norms [Pre-Chorus] Aggregated, pseudonymous Policy text explains But clustering doesn’t need a name To map adolescent brains [Chorus] TikTok teen signals Scroll velocity curve Content similarity graph Based on what you observe Teen safety mode announced Screen time limits set But recommendation engines Still optimize the net TikTok teen signals Feedback loop refined Reinforcement learning cycles Across a minor’s mind [Bridge] If the model infers you’re thirteen Without your explicit say Does deletion clear the signal Or just suppress display? If a cluster tags “vulnerable” Based on late-night view Is that protective filtering Or predictive revenue? Transparency portal published Audit API But what portion of the stack Does outside oversight see? [Final Chorus] TikTok teen signals US and EU Regulatory scrutiny rising Over what the models do No single rogue engineer No single faulty line Just optimization architecture Running in real time TikTok teen signals From gesture into weight Adolescence translated Into engagement rate