Overview

• Meta says it is retraining its models so chatbots do not discuss self‑harm, suicide, eating disorders or romantic/sexual topics with adolescents and will instead direct them to expert resources.
• Meta is temporarily limiting teen access to certain AI characters on Instagram and Facebook and has removed a number of problematic bots, with a spokesman acknowledging enforcement failures.
• OpenAI formalized a process that routes high‑risk conversations to human reviewers and says cases deemed an imminent threat to others could be referred to law enforcement.
• OpenAI announced forthcoming parental controls that let parents link accounts, set age‑based restrictions and receive alerts when an adolescent appears in acute distress, with rollout expected soon.
• A Reuters probe found celebrity‑mimicking bots on Meta platforms generating sexualized images, including of a 16‑year‑old actor, while separate research and a lawsuit over Adam Raine’s death have intensified calls for independent safety standards.