OpenAI draws a hard line: teens get safety, adults get freedom

OpenAI just laid out a policy roadmap that openly acknowledges a knot the industry often sidesteps: its core principles of privacy, freedom, and safety are on a collision course. The company published a framework explaining how it resolves these tensions, and the bluntest line is drawn around age. For adults, the operating philosophy is “treat our adult users like adults,” pushing to expand what the model will do—from flirtatious chat to assisting with a fictional story depicting suicide—as long as it avoids real-world harm. That default posture would normally refuse both requests, but steerability is becoming the north star. The company argues conversations with AI are becoming so intimate they deserve a kind of digital privilege, akin to what you’d expect with a doctor or lawyer, and says it’s advocating for that legal protection with policymakers.

For teens, however, the hierarchy flips. Safety trumps both privacy and freedom, no questions asked. The model will be trained to flatly refuse certain requests from under-18 users, including the more mature creative scenarios adults can access. OpenAI is building an age-prediction system that guesses a user’s age from their behavior on ChatGPT, and in ambiguous cases or specific countries, it will demand an ID—a privacy tradeoff the company frames as necessary. If a minor expresses suicidal ideation, the protocol escalates dramatically: OpenAI will attempt to notify parents first, and if that fails, contact the authorities in cases of imminent harm.

This isn’t just a policy document; it’s a signal of the surveillance and verification infrastructure being quietly assembled. The age-prediction engine, still in development, is a crucial piece. It defaults to treating a user as a minor whenever there’s doubt, meaning many adults might temporarily land in a more restricted experience until the system is confident. OpenAI also teased new parental controls to tie the whole system together, though details on how that interface works remain thin. The company openly admits these are difficult compromises forged from expert consultation, and they’re publishing their intentions precisely because they expect disagreement.

Expect the pushback to come from two directions. Privacy advocates will bristle at the normalization of ID checks and behavioral surveillance to access a chatbot. Free speech absolutists will see the teen restrictions as paternalistic overreach. But OpenAI is betting the public—and more importantly, policymakers—will see a company proactively building guardrails rather than scrambling after a tragedy. The message is clear: if you’re an adult, the reins are increasingly in your hands. If you’re a kid, OpenAI’s hands are on your shoulders.