OpenAI has introduced new safety features for under-18s using ChatGPT, including the ability to switch off human voice responses and set study mode hours. The company is also introducing regular reminders about ChatGPT being an AI tool.

The move follows growing concerns that people may believe AI tools are sentient. In a blog post, OpenAI emphasized the need for protections that reflect teenagers' developmental stage and support healthy use of the technology over time.

Teen accounts will become the default for those who identify as under 18, with specific prompts for harmful requests such as eating disorders also flagged to parents. Under-13s are still not allowed to use the product.

A recent report suggests that many UK universities may be at risk from students using AI in online exams, with 94% of students admitting to using tech in assessments.

The introduction of these features comes as there is currently no bespoke AI regulation in the UK, and similar restrictions have been considered for social media use by children. Other companies, including Anthropic and Google, also restrict AI tool access to adults only.