ChatGPT will guess if you’re a teen and start acting like a chaperone (www.techradar.com)

OpenAI is rolling out an age-guessing layer for ChatGPT that analyzes conversation patterns—questions, writing style, emoji use, how you respond to correction—to estimate whether a user is under 18. If the model suspects a teen (or is uncertain), it automatically switches the user into a filtered, teen-specific experience that restricts sexual content, gives more cautious responses around self-harm and mental health, and produces more “Sorry, I can’t help with that” refusals on sensitive topics. Adults who are misclassified must prove their age to regain the unrestricted experience; in ambiguous cases the system defaults to the safer mode. OpenAI also warns false positives will occur. This age-detection feature is part of a broader teen-safety push that adds parental controls—account linking, time limits, and alerts for detected “acute distress”—and the option to escalate serious cases to law enforcement if parents can’t be reached. Technically, the change puts behavioral and linguistic signals at the center of access control, shifting moderation from explicit opt-in settings to implicit classification. The implications are significant: it can reduce minors’ exposure to risky content but raises privacy, fairness, and autonomy concerns (misclassification, surveillance, chilling effects on adults with youthful styles), and concentrates high-stakes decisions in an imperfect classifier.