OpenAI has launched ChatGPT for Teens, introducing a set of safety features aimed at under-18 users of the AI chatbot. Among the changes, teen accounts will lose the ability to have ChatGPT respond in a human-sounding voice, a move intended to reduce the risk of the technology appearing lifelike, amid wider industry concern about people mistakenly believing AI tools are sentient. The update also adds break reminders and prompts telling teenagers "it can wait," while OpenAI insists the changes are not a response to any specific incident.

The new teen accounts will apply by default to users who identify as teenagers at sign-up, or whom OpenAI itself flags as under 18, though the company has not detailed how this identification will work; under-13s remain barred from the platform. Additional features include customisable "Study Mode" hours and "quiet times" when the tool switches off, alongside new alerts that notify parents, via a human-reviewed process within about an hour, if a teen makes eating disorder-related or other harmful requests. The launch comes as separate research found 94% of surveyed UK students admitted using AI in assessments, and follows criticism from the family of a teenager who died, who said OpenAI's earlier parental controls were insufficient.