OpenAI releases the Teen Safety Blueprint

OpenAI published the Teen Safety Blueprint, a short public document setting out principles it said should govern how AI chat products are built for minors, framed as a starting point for policymakers writing standards for the industry. It set out five guardrails: identifying likely minors through privacy-protective signals rather than hard identity checks; defaulting uncertain cases to a restricted under-18 experience, with an appeals path for adults wrongly flagged; blocking sexual, violent and self-harm content for that under-18 tier; giving parents configurable controls over a teenager’s account; and building in well-being tools and expert oversight, including a commitment that the product should not present itself as a substitute for therapy, friendship or an adult confidant.

The document did not carry new legal or technical commitments beyond what OpenAI had already begun rolling out; it functioned as a public statement of principles rather than a product announcement. It arrived one day after seven new wrongful-death and harm lawsuits were filed against the company over ChatGPT’s alleged role in psychological crises among teenagers and adults, and roughly two and a half months after the first such suit, brought by the parents of Adam Raine.

Cyberbullying Research Center co-director Sameer Hinduja called it “a solid baseline for the industry” but identified gaps, including that fewer than 1% of users had enabled parental controls under OpenAI’s existing opt-in system, that behavioural age-estimation might not catch minors using features such as voice and persistent memory that foster emotional attachment, and that the document offered no robust way to measure compulsive or escalating use.